next_group up previous


Bibliography

Torsten Müller,
Dresden University of Technology
Communications Laboratory, Chair for Telecommunications
WWW: http://www.ifn.et.tu-dresden.de/~muellert/
Email: mailto:muellert@ifn.et.tu-dresden.de

Note: This document contains pointers to literature and its content which seemed to be important to me. Therefore the given contents may not always reflect the real contents but my personal interests at a given time. Furthermore, it is partly in German and English.

Here is the bibtex file literatur.bib.


Contents

Sources of Information

Online-Bibliographies

QLD Dresden: http://www.ifn.et.tu-dresden.de/TK/deutsch/qldinter.htm
QLD Aachen: http://www.comnets.rwth-aachen.de/SFgate/cn-wais.html
Network Bibliography (H. Schulzrinne): http://www.cs.columbia.edu/~hgs/netbib/
CORA: http://www.cora.jprc.com
WEBBIB http://www.mcs.vuw.ac.nz/~cew/webbib/
DkDoc: http://www.cs.kau.se/~soren/dkdoc/papers.html

Definition of Terms




BegriffeNVT [BegriffeNVT]: "`Nachrichtenverkehrstheorie - Begriffe"'
ITG Recommendation 5.2-03, Entwurf 1997
Terms of traffic theory, in German and English, available at http://www.comnets.rwth-aachen.de/~itg/FG521/home_ger.html


dictionary.com [dictionary.com]: "`Dictionary.com"'
A online dictionary available at www.dictionary.com

DATACOM-lexicon: http://www.datacom-bookstore.de/lexikon/indexx.htm


Source Models / Traffic Characterization / Measurements




Crovella96 [Crovella96]: "`Self-Similarity in World Wide Web Traffic - Evidence and Possible Causes"'
investigates self-similarity in WWW traffic. A big measurement campaign with a modified browser is taken. The cause of self-similarity are heavy-tailed file sizes


ATMF970452 [ATMF970452]: "`Performance of Bursty World Wide Web (WWW) Sources over ABR"'
Usage of a http model with different classes of file sizes. See section 5.


PaxPhd97 [PaxPhd97]: "`Measurement and Analysis of End-to-End Internet Dynamics"'
Dissertation with 2 parts: 1. End-to-End Routing Behaviour, 2. End-to-End Internet Packet Dynamics. Extensive measurements in the Internet. Exact description of TCP behavior. Investigation of bandwidth, losses, delay.


Pax97 [Pax97]: "`End-to-End Internet Packet Dynamics"'
Results of extensive Internet measurements are analyzed. Measurements have been done at sender and receiver. A reduced version of: Paxson, PhD Dissertation, Berkeley University


Vica97 [Vica97]: "`Measurements and Modeling of WWW-Sessions"'
presents measurements of WWW sessions and WWW requests (pages) in an Ethernet segment. Plots the histogram of session size (mean=1.28MByte) and duration (mean=29 minutes) as well as page size (mean=54kByte) and inter response time (mean=81s). Inter response time and page size are assumed to be independent (scatter plot). A model with pareto distributed page size and inter response time is proposed and validated in an ATM simulation.


BMO+98 [BMO+98]: "`Multi-Layer Modelling of a Multi-Media Application"'
Analyzes the video teaching application ISABEL, see http://isabel.dit.upm.es/ Investigates several traffic models.


FPLZ98 [FPLZ98]: "`Measurements on Delay and Hop-Count of the Internet"'
A measurement series for delay and hop-count in the Internet is introduced which included about 3000 hosts. Traceroute has been used. It could be shown that no strong correlation exists between delay and hop-count. Very important is the country.


FRC98 [FRC98]: "`Efficient Policies for Carrying Web Traffic Over Flow-Switched Networks"'
A measurement of Internet traffic during 1 week has been conducted. The trace is analyzed. The paper gives distributions of packet size for different protocols (http, telnet, smtp) and its share. The size of flows has been analyzed (packets, bytes). Aim is the creation of short-cuts. Its overhead is analyzed. The influence of trigger for set up and timeout for set down has been investigated.


AZN98 [AZN98]: "`Broadband Traffic Modeling: Simple Solutions to Hard Problems"'
An overview on the problem of traffic modelling is given. It is shown that long range dependence is very important. The M/Pareto model for a small number of connections is introduced. Because of the central limit theorem it assumed that future calcuations can be done with a gaussian model. The influence of network protocols with feedback and congestion control (TCP) has not been a topic.


VBJ99 [VBJ99]: "`Analysis and Modeling of Traffic in Modern Data Communication Networks"'
Possion model, Traffic Train und Long-Range Dependence are introduced. The influence of LR dependence is shown with a curve of delay and utilization. LR dependence has a big influence. The packet train model is able to model the traffic of a single application. To model aggregated traffic, the Long-Memory model is better suited.


Schwieger99 [Schwieger99]: "`Messung der Charakteristik von TCP Verkehrsströmen in ATM-Netzen"'
Student thesis. The delay in real TCP end systems (PC, Workstation) to create TCP ACKS and Data has been measured. Furthermore, measurements of fairness have been conducted.


ST99 [ST99]: "`On Multimedia Networks: Self-Similar Traffic and Network Performance"'
Gives an overview about self-similarity and its influence on network performance. The reason of self-similarity are heavy-tailed file sizes. Its influence on performance is very strong. The larger the Hurst parameter, the smaller the performance and buffers have to be larger dimensioned.


Down99 [Down99]: "`Using pathchar to estimate Internet link characteristics"'
The tool pathchar (of Van Jacobson) has been introduced and evaluated. The tool serves to characterize Internet links on a path (delay, bandwidth, queueing delays). A technique has been introduced which reduces the time of measurement by 90 %.


AK99 [AK99]: "`On Traffic Models for TCP/IP"'
Long range dependence on the user level leads to long queues on the link level if no transport protocol is used. The article shows that this is not true if TCP is used as transport protocol. The user behavior is modeled as pareto distributed file size. The interarrival time is exponentially distributed. Furthermore, expo and deterministic file size have been used. If TCP is used only short queues can be observed. With measured traces this behavior can be verified. Therefore, Long-Range Dependence is not a real problem for TCP/IP traffic.


LHL99 [LHL99]: "`A Hierarchical Multi-Class Traffic Model for Data Networks"'
Based on traffic measurements a traffic model for Internet traffic has been developed. The overall traffic is divided into flows. Interarrival times as well as size of flows are log-gaus distributed. The IATs of packets are exponentially distributed or deterministic. By comparing the queueing performance of trace and model, the usability of the model is shown.


FBC99 [FBC99]: "`Statistical Evaluation and Modelling of Internet Dial-Up Traffic"'
Measurements (period of 6 months) in a ISDN pool have been analyzed. Holding and interarrival time are heavy-tailed and are depending on the time of day. The usage is following the pricing scheme. The hyperexpential distribution can be used to model IAT and holding time. If the exponential distribution is used instead, the blocking prob. is underestimated up to 1 magnitude.


AMI99 [AMI99]: "`Analysis of Internet Services in IP over ATM Networks"'
Measurement of Internet traffic at a university network (Spain) at a link to the Internet backbone (Dec. 1998, 1 week, 1.7e6 connection, 9e6 packets, 15GByte data): TCP 88%, Web 73%, number of bytes per connection, simultaneous number of connections, holding time of a connection, RTT, burst sizes.


YC99 [YC99]: "`Time series models for Internet data traffic"'
Analyzes network measurements between Ohio University and vBNS. 95% TCP, 2-5% UDP; 43%WWW, 20%FTP, 11% NNTP, ...


Char2000 [Char2000]: "`Internet Client Traffic Measurement and Characterisation Results"'
presents Internet traffic measurements in a modem (480000 flows) and 2.5 Mbit/s ADSL (1.5e6 flows) pool. Connections show significant idle prob. but can take the whole access bandwidth instantly. The upstream portion of the traffic is not neclectable. 15-20% of the connection even transmitted more traffic upstream. The fun factor has been introduced as measure for perceived QoS.


VKC00 [VKC00]: "`The Dependence of Internet User Traffic Characterization on Access Speed"'
Traffic Measurements (1999) of a Modem/ISDN Pool with 240 max. simultanious connections and an ADSL pool (free access) are analyzed. The measurement show strong influence of the tariffing structure. The volume and session duration increase with increasing access speed. Also the applications in use change. With ADSL the picture of the applications with respect to volume is completely changed compared to modem access. Only 12% is HTTP and 30% is FTP. The dominating application is a video appliacation with more than 55%. Also the average data volume for FTP connections is more than 3 MByte. The TCP connection data rate for ADSL is only increased by a factor of two compared to modem, which is very few compared to the potential increase of the access speed. Upstream and downstream traffic are not strongly correlated.


ML00 [ML00]: "`Variance of Aggregated Web Traffic"'
Real network traffic is burstier than any possible poisson traffic. The paper performs measurements and shows, that when aggregated, WWW traffic smoothes out as fast as poisson traffic. This means variance changes linearly with the mean bandwidth.


Plon00 [Plon00]: "`Internet Traffic Flow Size Analysis"'
This document contains the results of an informal investigation into the distribution of flow sizes between Fall 1999 and Spring 2000, ostensibly the period of time during which "file sharing" applications such as Napster, Gnutella, Scour Exchange, and the like, became more popular and generally available. This research is based upon internet traffic flow data collected at the University of Wisconsin - Madison. This document is available at http://net.doit.wisc.edu/data/flow/size/ Dave Plonka <plonka@doit.wisc.edu>


MC00 [MC00]: "`Trends in Wide Area IP Traffic Patterns: A View from Ames Internet eXchange"'
Reports measurement from a Internet traffic exchange point between big ISP networks. Measurements are performed from May 1999 to March 2000 and are based on port numbers. Traffic of new applications like Napster, IPSec and online gaming. The proportion of real-audio streaming and FTP (only at well known ports) has decreased. A big amount of traffic could not be associated with protocols (this could be e.g. FTP). The overall ratio between TCP and UDP has not changed.


Char00a [Char00a]: "`HTTP/TCP connection and flow characteristics"'
analyzes measurements of HTTP traffic. Gives distribution of flow duration, downstream volum (99.9 quantile is 1 Mbyte for port-to-port and 10 Mbyte for host-to-host) number of get-requests per flow, bitrate, and upstream to downstream bitrate ratio. This ratio is >1/10 for 80% of the flows and >1/1 for 30% of the flows showing that a large number of flows are about symetric.


Char00b [Char00b]: "`Web Performance in Practice - Why We are Waiting"'
Traffic Measurements have been performed at 2 ISP access networks. HTTP traffic has been deeply analyzed. The length of the phases of connections has been analyzed. Heavy tails have been found not only in the download time related to file size, but also in reaction times. Web performance is mainly influenced by DNS lookups.


Char00c [Char00c]: "`Internet Client Traffic Measurements and Characterization Results"'
Slides of COST-257 report. see http://www.jcho.de/jc/Pubs/ Describes results of a measurement campaign in 2 Internet access networks.


MH00 [MH00]: "`An Empirical Study of Real Audio Traffic"'
Presents results of a measurement of real-audio traffic from march 99 to june 99 at broadcast.com. Traffic differs significantly from www traffic. The mean flow size is 78 minutes, median is 45 min. 80% of traffic is from UDP, 20% TCP. Bitrate depends on contents, mostly 6.5, 16, and 20 kbit/s. Traffic is bursty, 6 packets on than a gap off (e.g. 1.8s). available at http://www.isi.edu/~johnh/PAPERS/Mena00a.html

Fairness




Jain84 [Jain84]: "`A Quantitative Measure of Fairness and Discrimination for Resource Allocation in Shared Computer System"'
Introduces the (non-weighted) fairness index and shows its features. A short description can be found in [ATMF940881] - ATMF940881.


ATMF990045 [ATMF990045]: "`Throughput Fairness Index: An Explanation"'
Explaines the (non-weighted) fairness index. The reasons for choosing this index in the ATM-Forum and its features are described.


VFJG99a [VFJG99a]: "`General Weighted Fairness and its Support in Explicit Rate Switch Algorithms"'
General Weighted Fairness is defined. Its usage in ERICA+ is shown. Simulations show, that ERICA+ is fair.


TCP

RFC's and Internet Drafts




RFC793 [RFC793]: "`RFC793: TRANSMISSION CONTROL PROTOCOL"'
RFC793, Standard document , specifies TCP


Pos80 [Pos80]: "`User Datagram Protocol"'
RFC 768, specifies User Datagram Protocol (UDP)


Bra89 [Bra89]: "`Requirements for Internet Hosts - Communication Layers"'
RFC 1122, This is one RFC of a pair that defines and discusses the requirements for Internet host software. This RFC covers the communications protocol layers: link layer, IP layer, and transport layer; its companion RFC-1123 covers the application and support protocols.


MR91 [MR91]: "`Gateway Congestion Control Survey"'
RFC 1254, gives an overview on scheduling and buffer management. Describes goals of congestion control including metrics like Power and Fairness. Describes congestion control: - source quench; - Random Drop for congestion recovery: a packet to discard is chosen from the queue, disadvantage for users with large RTT; - Random Drop for congesetion avoidance: discard arriving packets probabilistically. Gives no fixed implementation. Does not protect against misbehaving users. beat down problem; - DecBit scheme; - Selective Feedback: only feedback to users exceeding fair share; - Fair queueing: WFQ, SFQ (hashs different source and destination to different queues); - End system Congestion Control;


JBB92 [JBB92]: "`RFC1323: TCP Extensions for High Performance"'
presents TCP extensions to improve performance over large bandwidth*delay product paths and to provide reliable operation over high-speed paths. It defines new TCP options for scaled windows and timestamps, which are designed to provide compatible interworking with TCP's that do not implement the extensions. The timestamps are used for two distinct mechanisms: Round Trip Time Measurement and Protect Against Wrapped Sequences).


SCFJ96 [SCFJ96]: "`RTP: A Transport Protocol for Real-Time Applications"'
RFC 1889, defines RTP.


MMFR96 [MMFR96]: "`TCP Selective Acknowledgement Options"'
RFC 2018, defines the TCP SACK option. The algorithms for congestion control are not defined, but the standard algorithms have to be used.


AFP98 [AFP98]: "`RFC 2414: Increasing TCP's Initial Window"'
RFC 2414, proposes, to set the initial congestion window to 4 MSS.


RF99 [RF99]: "`A Proposal to add Explicit Congestion Notification (ECN) to IP"'
RFC2481, the ECN (explicit congestion notification) option to signal congestion to TCP sources instead of dropping packets is proposed.


APS99 [APS99]: "`TCP Congestion Control"'
RFC 2581, describes the algorithms for congestion control in TCP: Slow Start and Congestion Avoidance, Fast Retransmit/Fast Recovery, restarting idle connections, generating acks, and loss recovery mechanism.


FH99 [FH99]: "`The NewReno Modification to TCP's Fast Recovery Algorithm"'
RFC 2582, describes an algorithm for responding to partial acks referred to as NewReno.


HPF00 [HPF00]: "`TCP Congestion Window Validation"'
Defines, that TCP should use slow start after a long idle period. Cwnd and sstresh are halved each idle RTT. Also if the connection is application-limite, the cwnd should be decreased to midway between current cwnd and max. value used. Ssthresh has to be set to the max. of its current value and 3/4 of cwnd. Pseudocode is given, Simulations are performed with ns.


PA00 [PA00]: "`Computing TCP's Retransmission Timer"'
RFC 2988, This document defines the standard algorithm that Transmission Control Protocol (TCP) senders are required to use to compute and manage their retransmission timer. It expands on the discussion in section 4.2.3.1 of RFC 1122 and upgrades the requirement of supporting the algorithm from a SHOULD to a MUST. This document does not alter the behavior outlined in RFC 2581. The algorithm is the one described in [Jaco95].


Flo00 [Flo00]: "`Congestion Congrol Principles"'
RFC 2914, The goal of this document is to explain the need for congestion control in the Internet, and to discuss what constitutes correct congestion control. One specific goal is to illustrate the dangers of neglecting to apply proper congestion control. A second goal is to discuss the role of the IETF in standardizing new congestion control protocols.


ABF00 [ABF00]: "`Enhancing TCP's Loss Recovery Using Limited Transmit"'
RFC3042, This document proposes a new TCP mechanism that can be used to more effectively recover lost segments when a connection's congestion window is small, or when a large number of segments are lost in a single transmission window. The Limited Transmit algorithm calls for sending a new data segment in response to each of the first two duplicate acknowledgments that arrive at the sender. Transmitting these segments increases the probability that TCP can recover from a single lost segment using the fast retransmit algorithm, rather than using a costly retransmission timeout. Limited Transmit can be used both in conjunction with, and in the absence of, the TCP selective acknowledgment (SACK) mechanism [RFC2018].


SXMS00 [SXMS00]: "`Stream Control Transmission Protocol"'
RFC 2960, defines SCTP, another transport layer protocol which uses TCP congestion control.
SCTP is designed to transport PSTN signaling messages over IP networks, but is capable of broader applications. SCTP is a reliable transport protocol operating on top of a connectionless packet network such as IP. It offers the following services to its users: - acknowledged error-free non-duplicated transfer of user data, - data fragmentation to conform to discovered path MTU size, - sequenced delivery of user messages within multiple streams, with an option for order-of-arrival delivery of individual user messages, - optional bundling of multiple user messages into a single SCTP packet, and - network-level fault tolerance through supporting of multi- homing at either or both ends of an association. The design of SCTP includes appropriate congestion avoidance behavior and resistance to flooding and masquerade attacks.


Whet+01 [Whet+01]: "`Reliable Multicast Transport Building Blocks for One-to-Many Bulk-Data Transfer"'
RFC 3048, describes a framework for the standardization of bulk- data reliable multicast transport. It builds upon the experience gained during the deployment of several classes of contemporary reliable multicast transport, and attempts to pull out the commonalities between these classes of protocols into a number of building blocks. To that end, this document recommends that certain components that are common to multiple protocol classes be standardized as "building blocks". The remaining parts of the protocols, consisting of highly protocol specific, tightly intertwined functions, shall be designated as "protocol cores". Thus, each protocol can then be constructed by merging a "protocol core" with a number of "building blocks" which can be re-used across multiple protocols.

Articles etc.




FJ92 [FJ92]: "`On Traffic Phase Effects in Packet-Switched Gateways"'
Es wird beschrieben, was Phaseneffekte sind, und wie es zu diesen kommt. Es wird zunächst ein Szenario mit 2 Verbindungen und TCP/IP untersucht. Weiterhin werden ähnliche Effekte bei Decbit, größere TCP-Netzwerke und der Einfluß von Telnet Verkehr betrachtet. Random Drop Gateways zerstören Phaseneffekte.


Stevens94 [Stevens94]: "`TCP/IP Illustrated: the protocols"'
Very good book on TCP/IP. The "TCP bible". Illustrates TCP in detail.


Bakre94 [Bakre94]: "`I-TCP: Indirect TCP for Mobile Host"'
Eine Erweiterung für TCP für mobile Hosts wird vorgeschlagen und untersucht. Dabei wird die TCP-Schleife nicht bis zum Endsystem geführt, sondern am Mobile Switch aufgetrennt. Die Performancevorteile werden aufgetzeigt.


Jaco95 [Jaco95]: "`Congestion Avoidance and Control"'
Es werden (heute) wichtige Algorithmen in TCP vorgestellt und deren Notwendigkeit erklaert: RTT Schätzung, exponential backoff, slow-start, aggressive receiver ack policy, dynmaische Window-Anpassung bei Congestion. Ausgangspunkt der Untersuchung war ein dramatischer Leistungsrückgang im Internet. Das ursprüngliche Paper stammt aus dem Jahr 1988


RF95 [RF95]: "`Dynamics of TCP Traffic over ATM Networks"'
Es werden die Probleme bei TCP over plain ATM beschrieben. Eine vergleichende Simulationsstudie wird durchgeführt (ATM vs. Paketbetrieb). Early Packet Discard und Partial Packet Discard werden vorgestellt und bewertet.


KM95 [KM95]: "`Detailed Behavior of TCP over ATM"'
Shows, that TCP performs badly over plain ATM. The reason is, that ATM switches (without PPD or EPD) normally discard cells from multiple packets. TCP is not able to recover effectively from multiple losses without timeout. The authors propose to improve TCP to recover without timeout (They argue this is easier than using EPD).


BP95 [BP95]: "`TCP Vegas: End to End Congestion Avoidance on a Global Internet"'
Describes improvements of TCP congestion control realized in TCP Vegas and compares them to TCP Reno. Changes are done only at the sender. Retransmission is based on timestamps. Vegas checks, if there are segments to retransmit on every ack received (duplicate or normal). This avoids coarse-grained timeouts and only 1 duplicated ack is needed, which is effective for small CWND. CWND is decreased only, if the loss has been with the actual (and not an old CWND), this makes Vegas efficient for multiple losses per window. CWND is based on expected goodput, which is based on smalles measured RTT and the actual CWND and two thresholds alpha and beta. An experimental modification to Slow Start is proposed. Data segments are spaced based on an estimate of the bandwidth with a packet pair probing mechanism (similar to Keshavs).


Floyd95 [Floyd95]: "`TCP and Successive Fast Retransmits"'
Das Problem, daß bei TCP innerhalb eines Fensters 2mal Fast Retransmit durchgeführt wird, wird untersucht.


Manth95 [Manth95]: "`TCP Performance Over ATM Based LAN Interconnection Services"'
Simulative Untersuchung von TCP über SMDS(auf Basis von ATM bzw. DQDB). Das Problem stellt dabei die IWU dar. Um gute TCP-Performance zu errichen, muß deren Puffer groß sein, das führt zu unakzeptabler Verzögerung.


Mold95 [Mold95]: "`How a Large ATM MTU Causes Deadlocks in TCP Data Transfers"'
Es wird meßtechnisch TCP über ATM untersucht. Für bestimmte Kombinationen der TCP Sende- und Empfangspuffer kann es bei eingeschaltetem Nagle-Algorithmus zu Deadlocks kommen, die zu geringer Performance führen.


FF96 [FF96]: "`Simulation-based Comparisons of Tahoe, Reno, and SACK TCP"'
TCP Tahoe, Reno, NewReno, and SACK are explained. By simulation 1, 2, 3, and 4 packet losses are inserted. The transient behavior of the different TCP versions is observed. Reno TCP shows drastically decreased performance with more than 1 packet losses. TCP SACK has the best. Also available from http://www.aciri.org/floyd/papers.html


Loewe96 [Loewe96]: "`Abhängigkeit der Dienstgüteparameter bei Datenübertragung mittels TCPIP in ATM-Netzen"'
Diploma thesis, implements and describes TCP for YATS.


SHS97 [SHS97]: "`Characteristics of UDP Packet Loss: Effect of TCP Traffic"'
Die Charakteristik von UDP-Verlusten wird simulativ untersucht. Es wird ein WAN-Szenario mit TCP und UDP Verkehr betrachtet.Es wird eine Synchronisation der TCP-Verbindungen beobachtet. Diese hat negative Auswirkungen auf die UDP-Verbindungen. Es wird die Abhängigkeit von der Anzahl der Verbindungen und der UDP-Paketgröße untersucht.


MSMO97 [MSMO97]: "`The Macroscopic Behavior of the TCP Congestion Aviodance Algorithm"'
Ein Performance-Modell für den TCP Congestion Avoidance Algorithmus wird untersucht und durch Simulationen überprüft. Verschiedene TCP Implementationen unter verschiedenen Verlustbedingungen werden getestet.


Pax97b [Pax97b]: "`Automated Packet Trace Analysis of TCP Implementations"'
Describes an analysis of TCP traces measured in the Internet. The traces are automatically analyzed with a tool called tcpanaly. A big amount of traces produced by different TCP versions are considered. The analysis shows, that many TCP implementations are errorneous. Some are not according to the standards and can cause problems to network stability. The author argues that TCP is difficult to implement.


Allman98 [Allman98]: "`On the Generation and Use of TCP Acknowledgements"'
Verschiedene TCP-Acknowledgement-Strategien werden untersucht. Delayed Ack hat negative Eigenschaften bei Slow-Start.


BCCD+98 [BCCD+98]: "`Recommendations on Queue Management and Congestion Avoidance in the Internet"'
RFC 2309, points out the need to implement active queue management. RED is recommended as the standard buffer management algorithm. It provides reduced packet loss, lower delay, avoid lock-out behavior. The document deals with unresponsive and non TCP compatible protocols. As the amount of such traffic is expected to increase, the document points out, that this problem has to be addressed by future research. see also section 8.2.


JLN98 [JLN98]: "`A Detailed Experimental Performance Evaluation on TPC over UBR"'
Experimentell wird TCP über UBR untersucht. Es wird der Einfluß von EPD, PPD, Puffergröße, MTU-Größe, Fast Retransmit und Timergranularität betrachtet. Die Resultate zeigen gute Ergebnisse für TCP-Durchsatz und Fairness bei ausreichend großem Puffer.


WML98 [WML98]: "`Total Acknowledgements: A Robust Feedback Mechanism fo End-to-End Congestion Control"'
Shortpaper. Es wird eine Erweiterung der Acknowlegements in TCP vorgeschlagen. Neben dem empfangenen Byte werden auch die Anzahl weiterer empfangener Pakete gemeldet und für Congestion Control verwendet. Auch negative Acks sollen eingesetzt werden. Die Leistungssteigerung wird mittels Simulation gezeigt.


CO98 [CO98]: "`Differentiated End-to-End Internet Services using a Weighted Proportional Fair Sharing TCP"'
Es wird versucht, Weighted Proportional Fairness (WPF) unter TCP-Verbindungen zu erreichen. Das Gewicht soll dabei einstellbar sein und mit Billing bzw. Policing-Funktionen verknüpft werden. Es werden 2 Ansätze untersucht: 1) Anpassung des TCP Empfangs- puffers 2) Anpassung des TCP congestion avoidance algorithmus (MulTCP). Ziel von MulTCP ist es, sich wie mehrere unabhängige TCPs zu verhalten Die Eigenschaften werden simulativ untersucht. Im Anhang wird die Definition von max-min Fairness, proportional Fairness und WPF vorgenommen. Es existieren Literaturverweise auf das Thema Fairness.


KVR98 [KVR98]: "`Two-Way TCP Traffic over Rate Controlled Channels: Effects and Analysis"'
Bidirektionales TCP wird untersucht. Dabei wird auf die Wirkung von ACK compression eingegangen. Dadurch werden mehrere Acks nacheinander versendet, welches zu einem stark bursthaften Verhalten führt. Das wiederum führt zu einer schlechten Auslastung des Links. Auch bei ABR kann Ack compr. beobachtet werden, allerdings verursacht durch die Endsysteme. Es erfolgt eine gute Beschreibung des dynamischen Verhaltens. Es werden Berechnungen für symmetrische und asymmetrische Bandbreite vorgenommen.


Theo98 [Theo98]: "`ACC: Using Active Networking to Enhance Feedback Congestion Control Mechanisms"'
Es wird Active Congestion Control (ACC) vorgestellt, welches auf Aktiver Netzwerktechnologie beruht. Bei Überlast werden dabei im vorhergehenden Router Paketfilter installiert, welche Pakete verwerfen und das TCP-Windowupdate durchführen. Das System wird mit ns simulativ untersucht, wobei Drop Tail FIFO und RED zum Einsatz kommt. Bei hohen Bandbreite-Verzögerungsprodukten wird ein Durchsatzgewinn von etwa 18 Prozent erreicht.


KLS98 [KLS98]: "`Beyond Best Effort: Router Architectures for the Differentiated Services of Tomorrow's Internet"'
stellt Router-Architektur vor. Geht auf Fair Queueing ein. Das ist schwierig, da sehr viele (TCP) Flows (256000 bei OC-3) auftreten können. Deshalb wird FQ für Klassen durchgeführt. Router müssen TCP unterstützen (Bursthaft, Unfair, Synchronization, Random Loss Sensitivity). Per Flow Queuing bzw. Puffer Management allein sind nicht ausreichend. Eine Kombination FQ-ALQD (approximated longest queue drop) wird gegen FQ-RED und FCFS-RED verglichen und erreicht wesentlich bessere Ergebnisse bei extensivem CBR-Verkehr (UDP).
siehe auch Abschn. 8.2.


LK98 [LK98]: "`TCP Fast Recovery Strategiew: Analysis and Improvements"'
The paper investigates TCP performance problems and suggests improvements. TCP Timeouts are mainly caused by small congestion windows, which is shown by traces. The proposed solution is to allow the sender to inject two new packets on receiving duplicated acks and adjust the cwnd accordingly. Another problem is the cwnd after multiple losses. In NewReno and SACK the cwnd is only halved, which is too aggressive. Other optimizations are proposed including one to recover from lost retransmissions. The changes are verified by simulations and measurements.


PFTK98 [PFTK98]: "`Modeling TCP Throughput: A Simple Model and its Empirical Valiation"'
An analytical model for TCP throughput is developed and validated. The model includes losses by Fast Retransmit as well as timeouts. Measurements (TCP traces) show, that including timeouts is important because they influence performance severly and occur more often than Fast Retransmits. The model can predict the performance over a wide range of loss rates.


Schueler98 [Schueler98]: "`Fairness bei TCP-Datenübertragung in ATM-Netzwerken"'
Diploma thesis. Investigates fairness issues using TCP/IP in ATM networks for loss-free and loss systems. Uses YATS.


MS99 [MS99]: "`Traffic Phase Effects in Realistic IP Networks"'
investigates phase effects.


SM99 [SM99]: "`Dynamics of TCP in congested ATM multiplexers"'
invetigates TCP over ATM for EPD and drop tail. Explains effects in detail.


BSJ99 [BSJ99]: "`Quality of Service using Traffic Engineering over MPLS: An Analysis"'
Einfache Simulationsstudie. TCP und UDP Verkehrsströme werden transportiert. Werden TCP und UDP in verschiedenen Trunks transportiert, und erfolgt CBQ-Scheduling, ergibt sich gute Performance, sonst unfaires Verhalten.
siehe auch Abschnitt 7


TTL99 [TTL99]: "`The Challenges of Seamless Handover in Future Mobile Multimedia Networks"'
enthält u.a. Verbesserungen für TCP in Mobilen Netzen


FF99 [FF99]: "`Promoting the Use of End-to-End Congestion Control in the Internet"'
Durch nicht-TCP-Anwendungen wird congestion avoidance and control im Internet gefährdet (Collapse). Das wird durch einfache Simulationen gezeigt. Dieser Artikel schlägt vor, nicht-TCP-friendly Flows zu erkennen und deren Bandbreite zu begrenzen. Unterschieden werden: TCP-friendly, unresponsive, flows using disproportionate bandwidth. Alternative Ansätze (Pricing, Scheduling) werden diskutiert, aber als nicht ausreichen abgelehnt.


PKT99 [PKT99]: "`A TCP-Friendly Rade Adjustment Protocol for Contiuous Media Flows over Best Effort Networks"'
Es wird vorgeschlagen, für CM (continuous media) Applikationen eine Steuerung gemäß TCP vorzunehmen. TCP selbst ist für CM nicht anwendbar, da bei TCP Fehlersteuerung und Congestion Control gekoppelt sind, CM aber oft fehlertolerant ist. Es wird durch Simulationen und Messungen am realen System gezeigt, daß das entwickelte Protokoll CMTCP sich ähnlich wie TCP verhält.


Ghani99 [Ghani99]: "`Enhanced TCP/IP ACK Pacing for High Speed Networks"'
Es wird ein Algorithmus zu Regelung von TCP-Verbindugnen vorgestellt, der darauf beruht, die TCP-Acks zu verzögern. Dabei wird der erreichte Durchsatz in die Vorwärtsrichtung gemessen. Der Algorithmus ist einfach zu parametrisieren und robust. Die Untersuchung erfolgte and 2 Szenarien simulativ.


YR99 [YR99]: "`Modeling TCP Behavior in a Differentiated Services network"'
siehe auch 6


MR99 [MR99]: "`Arguments in favour of admission control for TCP flows"'
Es wird CAC für elastischen Verkehr (TCP-Verbindungen) vorgeschlagen. Die negativen Auswirkungen, wenn man keine CAC durchführt werden analytisch gezeigt (die Modelle leiden aber unter den für analytische Berechnungen notwendigen Vereinfachungen). Die CAC besteht darin, die Anzahl von TCP-Verbindungen zu begrenzen. Realisationsmöglichkeiten werden beleuchtet.


Feng99 [Feng99]: "`Improving Internet Congestion Control and Queue Management Algorithms"'
Die Dissertation schlägt Verbesserungen des Congestion Controls im Internet vor und untersucht diese mittels Simulation und Messungen. Es werden existierende Techniken vorgestellt. BLUE (und Erweiterungen), eine Verbeserung für RED wird vorgeschlagen und vergleichend untersucht. TCP in einer DiffServ-Umgebung wird untersucht. Packet-Marking für TCP und nicht-TCP Ströme wird untersucht.


CG99 [CG99]: "`QoS Control and Interworking of Overbooked Elastic and Brittle ATM Traffic"'
siehe Abschn. 9


BPK99 [BPK99]: "`The Effects of Asymmetry on TCP Performance"'
Die Auswirkungen von Netzwerk-Asymmetrie, wie sie in mobilen Netzwerken auftreten, werden untersucht und Verbesserungen vorgeschlagen (Techniken zur Verringerung der Rate der Acks, Verringerung der Burstiness bei infrequenten Acks und unterschiedliche von Daten und Acks beim Scheduling). Es werden Messungen und Simulationen durchgeführt.


YR99a [YR99a]: "`Realizing throughput guarantees in a differentiated services network"'
Es werden Techniken untersucht, um QoS-Garantien in DiffServ zu ermöglichen: Limitierung der OUT-Pakete, Einsatz von 3 drop precedence level (Schutz UDP), Markierung entsprechend gewünschter Rate (TCP-friendly), Modifizierung von TCP (2 Fenster, guaranteed, best effort Fenster), Modifizierung der Markierstrategie, Modifizierung der Drop policie am Router. Der Einfluss von nonresponsive Flows und RTT wird untersucht.


YR99b [YR99b]: "`Marking for QoS Improvement"'
Untersucht Performance bei TCP über DiffServ. Mechanismen zur Markierung von Paketen werden vorgeschlagen: IN-fair und BW-fair Marking. Simulationen zeigen, das diese sich besser verhalten als Proportional Marking. Weiterhin wird Adaptive Marking vorgeschlagen. Dabei wird die Marking-Rate dynamisch schrittweise verändert. Weiterhin wird ein Receiver-Controlled Scheme vorgeschlagen, welches auf dem ECN-bit beruht, und die aktuell erreichte Rate am Receiver mit dem Sollwert vergleicht. available at: http://ee.tamu.edu/~reddy/


YR99c [YR99c]: "`Impact of marking strategy on aggregated flows in a differentiated-services network"'
Introduces proportional marking, IN-fair marking, and BW-fair marking. IN-fair and BW-fair marking are able to improve fairness among different TCP flows.


GD99 [GD99]: "`TCP/IP Enhancements for Satellite Networks"'
beschreibt und untersucht Erweiterungen für Satellitennetzwerke und TCP-Verkehr (high bandwidth delay product): Auswirkung Verluste, Delay, Asymetrie des Hin und Rückkanal. Verbesserungen: FEC, ARQ; Verbesserungen TCP: vergrößertes Fenster, PAWS, Timestamp (TCP echo option) (RFC 1072), Daten bei Verbindungsaufbau (RFC 1644), byte counting, MTU discovery, TCP Vegas (delay sensitive), NewReno, SACK (RFC 2018), FACK (entkoppelt congestion control und data recovery), drop from front (für Acks), Ack filtering, NACK, multiple TCP sessions. Genauer werden mehrfache TCP-Verbindungen und ACK Control Schemes untersucht.


CFE99 [CFE99]: "`Performance Evaluation of TCP Extensions on ATM over High Bandwidth Delay Product Networks"'
untersucht experimentell Performance von TCP ueber ATM Satellitennetzwerk. TCP-Versionen: Reno, NewReno, SACK. Bei Verlusten koennen alle Versionen keinen guten Durchsatz erzielen, SACK ist am besten. Bei Übelast kann Shaping Vorteile bringen. TCP SACK ist auch hier am besten. TCP SACK ist aggressiver gegenüber anderen TCP Verbindungen.


Sama99 [Sama99]: "`Return Link Optimization for Internet Service Provision Using DVB-S Networks"'
untersucht Performance von TCP ueber Satellitennetzwerk, wobei vor allen Dingen auf Asymtetrie eingegangen wird. Es werden Header Compression, Ack supression (Ausfiltern von Acks), Ack Compaction und Ack Spacing untersucht. Bei Ack compaction werden acks zunächst gelöscht, um Kapazität zu sparen, dann aber wieder eingefügt. Bei Spacing wird zusätzlich noch geshapt. Ack Compaction und Ack Spacing erreichen gute Performance. Spacing verhindert Ack compression effect. Dadurch wird der Pufferbedarf deutlich verringert.


LJ99 [LJ99]: "`Improving Explicit Congestion Notification with the Mark-Front Strategy"'
Es wird Explicit Congestion Notification untersucht. Dabei wird bei Mark-Front eine bessere Performance als bei Mark-tail (utilization, fairness). Der Grund ist, dass TCP Fast Retransmit schneller ueber verlorene Daten informiert wird. Dadurch wird weniger Puffer benötigt. Es wird RED benutzt. Auch neue Verbindungen profitieren von Mark-Front. siehe auch Abschn. 8.2


SLSC99 [SLSC99]: "`Buffer Management Schemes for Supporting TCP in Gigabit Routers with Per-Flow Queueing"'
es wird untersucht, welche Auswirkung Scheduling bzw. Buffer Management auf TCP-Flüsse hat. Es werden FCFS und FQ bzw. Tail Drop, RED, LQD (Longest Queue Drop), RND (Auswahl zufälliger Backlogged Flüsse mit mehr als fair Share) und AQLD (Approx. LQD) betrachtet (LQD, ALQD und RND sind Drop from Front). Scheduling allein reicht nicht aus, um Fairness (bzw. Garantien) zu erreichen. Tail Drop hat die schlechtesten Eigenschaften, Drop from Front Strategien sind prinzipiell günstiger, da Fast Retransmit schnell informiert wird. Auch RED ist ungünstig, da kein Wissen über Flows existiert. Es treten Mehrfachverluste auf und wiederholte Pakete werden verworfen. Unter den anderen Verfahren ist FQ-LQD vorzuziehen. Siehe auch Abschn. 4.


Sun99 [Sun99]: "`Solaris 7 System Administrator Collection: TCP/IP and Data Communications Administration Guide"'
Online-Manual, describes (among many other things) how to use Window Scale Option and TCP SACK in Solaris 7.


AF99 [AF99]: "`On the Effective Evaluation of TCP"'
gives advise, how to do performance evaluation of TCP. It starts from the fact, that many research on TCP is misleading. To get useful results one should: - use simulation, tests and live internet test; - use basic TCP alg. as well well-known improvemetns as SACK, Nagle and delayed ack; - window large enough; - Data should be FTP as well as HTTP and other realistic (competing, non-TCP) traffic; - useful MTU size;


SCWA99 [SCWA99]: "`TCP Congestion Control with a Misbehaving Receiver"'
The article shows, how a TCP receiver can cause the TCP sender to not use TCP congestion control. The methods are: 1. ACK division where many ACKs are sent with increasing sequeuence numbers; 2. DupACK spoofing where many duplicated Acks are sent; 3. Optimistic ACKing where data are acked before actually receiving the data; All methods show to "improve" the goodput of the receiver using it. The paper shows, that many real TCP senders are vulnerable to such attacks and gives hints how to fix the problem.


All99 [All99]: "`TCP Byte Counting Refinements"'
An improved byte counting algorithm is proposed which overcomes the problem of limited number of ACKs in the delayed ACK. It is called ABC (Appropriate Byte Counting) and improves LBC (Limited Byte Counting). LBC is too aggressive. ABC uses the LBC algorithm only during the initial slow start phase and is less aggressive. Another scaled version is proposed, which allows to tune aggressiveness. The impact on other flows is investigated.


AP99 [AP99]: "`On Estimating End-to-End Network Path Properties"'
Investigates the influence of TCP RTO estimate and bandwidth estimate for correct ssthresh setting by analyzing TCP traces. Explains RTT measurement. is typically 2 Ticks = 2*500ms. has big influence on performance. Also bad retransmissions have big influence, because they lead to decreased CWND and the RTT estimate is wrong. After detecting a bad retransmission, the changes could be taken back. For estimating the bandwidth, packet pair does not work well. Sender-side algorithms in general are problematic because the ACK stream spacing may not be preserved by the network.


BAD00 [BAD00]: "`On TCP performance in a Heterogeneous Network: A Survey"'
Surveys some aspects of TCP performance. Versions: Discusses Fast Recovery in Reno, NewReno, SACK. Buffer size requirements: to avoid timeouts during SS is mue*T/2, to reach full utilization during Cong. Avoid. mue*T. RTT: bias against long RTT, discusses larger initial windows, byte counting, limited byte counting to avoid large bursts in case of Ack gaps, TCP spoofing, fairness (FRED). Noncongestion Losses. Bandwidth asymetry: ack losses, increasing RTT, ack compression, grow of congestion window, header compression, reducing ack rate, ack substitution (reconstruct a lost ack)


BPS99 [BPS99]: "`Packet Reordering is Not Pathological Network Behavior"'
presents measurements which show, that packet reordering is common behavior in current Internet and not only caused by malfunctions. It happens especially for higher loads and leads to poor TCP performance. The reason is parallelism to increase performance. Reordering is hardly to avoid, possible solutions are discussed (switch design, TCP changes).


LK00 [LK00]: "`The Eifel Algorithm: Making TCP Robust Against Spurious Retransmissions"'
In mobile systems disconnections can occur in the order of seconds which can lead to sudden increase in RTT. After an (unnecessary) retransmission (spurious timeout) the sender does not know, wheter it receives an ACK for the original or retransmitted packet. Because of TCP congestion control the sender reduces its load. (Similar problems occur due to reordering called spurious fast retransmit.) An algorithm called Eifel is proposed. It discriminates between original and retransmitted packet by using the timestamp option.


ABR / Flow Control / Feedback (außer TCP)




Jain90 [Jain90]: "`Congestion Control in Computer Networks: Issues and Trends"'
Es wird der Mythos widerlegt, daß genuegend Bandbreite, Prozessorkapazität oder Speicher ausreicht, um Congestion zu vermeiden. Es werden die Grundprinzipien von Congestion Management erläutert und warum das Problem schwierig ist. Es wird gezeigt, daß verschiedene Ursachen von Congestion verschiedene Lösungsansätze benötigen. Einige Ansätze werden genauer vorgestellt.


Jain92 [Jain92]: "`Myths About Congestion Management in High-Speed Networks"'
Es werden bestimmte Aussagen über Congestion Management diskutiert. Vergleichend wird eingegangen auf: reservation/walk-in, open-loop/feedback control, rate control/window control, router based/source based control. Es wird gezeigt, dass Backpressure nur unter bestimmten Voraussetzungen sinnvoll ist und dass i.A. eine Kombination von Congestion Managment Schemen benoetigt wird.


Kori92 [Kori92]: "`Why is Flow Control Hard: Optimality, Fairness, Partial and Delayed Information"'
Das Themengebiet Flow Control wird überblicksartig präsentiert. Es werden die Probleme dargestellt. Verfahren werden vorgestellt: Decbit, TCP, Rate-Based. Optimale Flow Control wird betrachtet, Power und QoS. Dem Thema Fairness und Delayed Information werden eingene Kapitel gewidmet.


OSV94 [OSV94]: "`Reliable and Efficient Hop-by-Hop Flow Control"'
Flußkontrollschema auf Credit-Basis wird beschrieben. Basiert auf Kredit für Puffer, kann aber auf Ratenschema ausgedehnt werden.


KBC94 [KBC94]: "`Credit-Based Flow Control for ATM Networks: Credit Update Protocol, Adaptive Credit Allocation, and Statistical Multiplexing"'
Ein kreditbasiertes Flußkontrollschema wird präsentiert.


LSTIS95 [LSTIS95]: "`TCP Performance over ABR and UBR Services in ATM"'
Der Transport von TCP mit ABR und UBR (mit EPD) über ATM wurden simulativ untersucht. UBR zeigt schlechte Performance bzw. braucht große Puffer, ABR funktioniert besser, hat auch bessere Fairness. VC queueing/accounting kann bei UBR die Leistung verbessern.


ST95 [ST95]: "`Intelligent Congestion Control for ABR Service in ATM Networks"'
Verschiedene Congestion Control Algorithmen werden vorgestellt. Das Ratenbasierte Schema intelligent congestion control wird näher untersucht.


YR95 [YR95]: "`A Taxonomy for Congestion Control Algorithms in Packet Switching Networks"'
Eine neue Ordnung für Congestion Control Algorithmen wird präsentiert. Grundlegende Probleme des Congestion Control werden erörtert. 22 Algorithmen werden kurz vorgestellt und eingeordnet. Referenzen führen zu den Congestion Control Algorithmen


ERICA96 [ERICA96]: "`ERICA Switch Algorithm: A Complete Description"'
Der ABR-Switchmechanismus ERICA (und ERICA+) wird detailliert diskutiert. Die Gründe für die einzelnen Teilmechanismen werden angegeben. Der Pseudocode ist enthalten.


Lefel96 [Lefel96]: "`Congestion Control for Best-Effort Service: Why We Need a New Paradigm"'
Es werden die Probleme des Congestion Control im Internet vorgestellt. Die Probleme mit TCP erden erläutert. Es wird eine neue Herangehensweise bestehend aus Scheduling, Buffer Management, Feedback und End Adjustment vorgeschlagen. Beispiele verschiedener Verfahren komplettieren den Beitrag.


Manth97 [Manth97]: "`A comparison of ABR and UBR to support TCP traffic"'
Die Performance von ABR und UBR wird verglichen. UBR zeigt hohen Durchsatz wenn genügend Puffer vorhanden ist. Die Komplexität von ABR kann keine Verbesserung erreichen. Bei kleinen Puffern jedoch zeigt ABR bessere Performance. ABR zeigt ein besseres Fairnessverhalten.


Roberts97 [Roberts97]: "`Explicit Rate Flow Control - A 100 Fold Improvement over TCP"'
verbesserte Version des ATMForum-Dokuments 94-0448; Erklärt die Vorteile von Rate Control gegenüber TCP


Ott97 [Ott97]: "`TCP over ATM: ABR or UBR?"'
Simulationsstudie untersucht TCP über ATM (ABR und UBR). Dabei zeigt sich, dass ABR keine bessere Leistungsfähigkeit als UBR aufweist.


Bau97 [Bau97]: "`Einige Gedanken und Ergebnisse zu TCP über UBR/EPD/ABR"'
Simulationsstudie, durchgeführt mit YATS, untersucht TCP über UBR, UBR+ und ABR. Ohne Hintergrundverkehr zeigt UBR+ die besten Ergebnisse, ABR ist am fairsten. UBR kann keine gute Performance erzielen. Mit HG-Verkehr erzielt ABR die schlechtesten Ergebnisse, da es die Pausen nicht effektiv nutzen kann.


ATMF970452 [ATMF970452]: "`Performance of Bursty World Wide Web (WWW) Sources over ABR"'
WWW server and clients are modelled, are transmitted over ABR. Open TCP connections are used, therefore, TCP slow start cannot work which leads to bursts. ABR can work well with these bursts. Usage of a http model with different classes of file sizes. See section 2.


ATMF980290 [ATMF980290]: "`Improving TCP/IP Performance over ATM's ABR Service with Explicit Window Adaption"'
Beschreibt die Anpassung von ABR und TCP. Die Flußsteuerung wird verbunden über das WND-Feld der Acknowledgements. Dadurch wird Überlast im Edge-Router vermieden. Simulationsresultate sind angegeben.


ATMF980152 [ATMF980152]: "`Explicit rate control of TCP applications"'
Beschreibt die Anpassung von ABR und TCP. Die Flußsteuerung wird verbunden über das WND-Feld der Acknowledgements. Dadurch wird Überlast im Edge-Router vermieden. ABR verwendet Erica. Simulationsresultate sind angegeben.


Fahm98 [Fahm98]: "`ABR Flow Control for Multipoint Connections"'
Point-to-multipoint und multipoint-to-point flow control für ABR service wird untersucht.


Siegel98 [Siegel98]: "`Rate Control to Enhance Network Service Quality"'
Die Wichtigkeit von Rate Control im Internet wird unterstützt. Es wird erklärt, warum QoS wichtig ist, grundlegende TCP-Mechanismen erklärt und warum diese nicht ausreichend sind.


KL98 [KL98]: "`Performance Analysis of a Rate-Based Feedback Control Scheme"'
Feedback Control System (ABR) mit einem Knoten und Quellen wird analytisch untersucht. Es kommt der Quasi-birth-death Prozeß zum Einsatz. Es wurden feedback delay und Quellenperiodenzeiten untersucht. Das Ratenbasierte System verhält sich besser als binäres Feedback.


Theo98 [Theo98]: "`ACC: Using Active Networking to Enhance Feedback Congestion Control Mechanisms"'
siehe auch TCP


VFJG99b [VFJG99b]: "`QoS and Multipoint Support for Multimedia Applications over the ATM ABR Service"'
Es wird gezeigt, daß ABR für den Transport von Multimedia passend ist. Es ist passend, da ABR sowohl kleine Queues bildet, als auch eine MCR und geringe Zellverluste garantiert. Videoanwendungen können Congestion-Signale vom Netzwerk auswerten (z.B. layered coding). Auch Mulitpoint connections sind bei ABR möglich.


ATMF990403 [ATMF990403]: "`A Framework for Virtual Channel onto Virtual Path Multiplexing in ATM-ABR"'
Es wird ein Framework vorgestellt, um ABR-VCs in ABR-VPs multiplexen zu können. Dabei wird zwischen Netzwerken mit/ohne VS/VD unterschieden. Es werden verschiedene Fairness-Definitionen vorgestellt. VPC/VCC Erica+ wird simulativ untersucht.


Costeux99 [Costeux99]: "`Fluid Analysis of a TCP Connection over ABR"'
Mittels Fluid-Flow Methode wird eine TCP-Verbindung (Tahoe) über ABR (EFCI) modelliert und analysiert. Dazu wird das Verhalten in verschiedene Phasen eingeteilt, und für diese Phasen werden Differentialgleichungen aufgestellt.


MG99 [MG99]: "`Intelligent Credit Management for a Credit Based Congestion Control Procedure"'
Es wird ein kreditbasiertes Flußregelungsverfahren für das System X.n vorgeschlagen. Dieses erlaubt es den Sendern, auch ohne vorhandene Kredits zu senden, wobei evtl. Verluste auftreten können, die zu Retransmissions führen. Die Leistungsfähigkeit wird simulativ untersucht. Das Verfahren führt bei kleinen Puffern (etwa RTT) zu guten Ergebnissen.


LB00 [LB00]: "`Pathological Behaviors for RLM and RLC"'
Shows problems with RLM and RLC. A new mechanism called PLM which can however only work with fair queueing is proposed.


DiffServ / IntServ




BCS94 [BCS94]: "`Integrated Services in the Internet Architecture: an Overview"'
RFC 1633 (informational), defines IntServ, Abstract: This memo discusses a proposed extension to the Internet architecture and protocols to provide integrated services, i.e., to support real- time as well as the current non-real-time service of IP. This extension is necessary to meet the growing need for real-time service for a variety of new applications, including teleconferencing, remote seminars, telescience, and distributed simulation.


BBCD+98 [BBCD+98]: "`An Architecture for Differentiated Services"'
RFC 2475. defines the DiffServ architecture and terminology.


HBWW99 [HBWW99]: "`Assured Forwarding PHB Group"'
RFC 2597, defines Assured Forwarding (AF) PHB group


JNP99 [JNP99]: "`An Expedited Forwarding PHB"'
RFC 2598, defines EF PHB. Provides simulation results.


NBBB98 [NBBB98]: "`Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers"'
RFC 2474, standard track, defines the DS-field of IPv4 and IPv6.


GB98 [GB98]: "`Interoperation of Controlled-Load Service and Guaranteed Service with ATM"'
Specifies the Interworking between IntServ and ATM. Specifies CBR or rt-VBR for GS and CBR, nrt-VBR or ABR for CLS. Excess traffic can be sent in CLS and is handled by setting CLP=1.


SW97 [SW97]: "`Network Element Service Specification Template"'
RFC 2216, defines framework for specifying new IntServ services
This document defines a framework for specifying services provided by network elements, and available to applications, in an internetwork which offers multiple qualities of service. The document first provides some necessary context - including relevant definitions and suggested data formats - and then specifies a "template" which service specification documents should follow. The specification template includes per-element requirements such as the service's packet handling behavior, parameters required and made available by the service, traffic specification and policing requirements, and traffic ordering relationships. It also includes evaluation criteria for elements providing the service, and examples of how the service might be implemented (by network elements) and used (by applications).


Wroc97 [Wroc97]: "`Specification of the Controlled-Load Network Element Service"'
RFC 2211, specifies Controlled Load Service


SPG97 [SPG97]: "`Specification of Guaranteed Quality of Service"'
RFC 2212, defines IntServ Guaranteed QoS


BZB+97 [BZB+97]: "`Resource ReSerVation Protocol (RSVP) - Version 1 Functional Specification"'
RFC 2205, defines RSVP


Cra+98 [Cra+98]: "`A Framework for Integrated Services and RSVP over ATM"'
Informational RFC 2382, provides a framework for Integrated Services and RSVP over ATM


HG99a [HG99a]: "`A Single Rate Three Color Marker"'
RFC 2697, This document defines a Single Rate Three Color Marker (srTCM), which can be used as component in a Diffserv traffic conditioner [RFC2475, RFC2474]. The srTCM meters a traffic stream and marks its packets according to three traffic parameters, Committed Information Rate (CIR), Committed Burst Size (CBS), and Excess Burst Size (EBS), to be either green, yellow, or red. A packet is marked green if it doesn't exceed the CBS, yellow if it does exceed the CBS, but not the EBS, and red otherwise.


HG99b [HG99b]: "`A Two Rate Three Color Marker"'
RFC 2698, this document defines a Two Rate Three Color Marker (trTCM), which can be used as a component in a Diffserv traffic conditioner [RFC2475, RFC2474]. The trTCM meters an IP packet stream and marks its packets based on two rates, Peak Information Rate (PIR) and Committed Information Rate (CIR), and their associated burst sizes to be either green, yellow, or red. A packet is marked red if it exceeds the PIR. Otherwise it is marked either yellow or green depending on whether it exceeds or doesn't exceed the CIR.


FSN00 [FSN00]: "`A Time Sliding Window Three Colour Marker (TSWTCM)"'
RFC 2859, this memo defines a Time Sliding Window Three Colour Marker (TSWTCM), which can be used as a component in a Diff-Serv traffic conditioner [RFC2475, RFC2474]. The marker is intended to mark packets that will be treated by the Assured Forwarding (AF) Per Hop Behaviour (PHB) [AFPHB] in downstream routers. The TSWTCM meters a traffic stream and marks packets to be either green, yellow or red based on the measured throughput relative to two specified rates: Committed Target Rate (CTR) and Peak Target Rate (PTR).


SNH00 [SNH00]: "`An Assured Rate Per-Domain Behaviour for Differentiated Services"'
Internet Draft draft-ietf-seddigh-pdb-ar-00.txt, defines the Assured Rate Per-Domain Behavior, this is a service with minimum rate guarantee (CIR), allowing excess bitrate up to PIR with a burst size of CBS. The AR PDB is suitable for carrying trafficaggregates that require rate assurance but do not require delay and jitter bounds. It can be created using the diffserv AF PHB along with suitable policers at the domain ingress nodes.


Nic01 [Nic01]: "`Definition of Differentiated Services Per Domain Behaviors and Rules for their Specification"'
Internet Draft draft-ietf-diffserv-pdb-def-03, defines how to construct Per-Domain Behaviors out of PHBs.




ABF97 [ABF97]: "`Scalable Resource Reservation fot the Internet"'
Es wird eine skalierbare Reservierungsstruktur für das Internet vorgestellt, die darauf basiert, daß die Endsysteme die Reservierungen verwalten. Die Endsysteme senden request oder reservation-Pakete und bekommen von den Routern Bandbreite zugeteilt. Nach einem Timeout erlischt die Reservierung. Durch ein Feedbackprotokoll erfährt der Sender, ob die Reservierung erfolgreich war. Simulationen zeigen die Leistungsfähigkeit des Ansatzes.


NJZ97 [NJZ97]: "`A Two-bit Differentiated Services Architecture for the Internet"'
Eine Architektur für DiffServ wird präsentiert. Diese basiert auf Assured Service (AF) und Premium Service (EF), welche gemeinsam verwendet werden. Die benötigten Mechanismen wie Klassifizieren, Markieren, Policing, Shapen, Prioritäten etc. werden vorgestellt. Es werden Szenarien entworfen, wie der Dienst genutzt werden kann. Dazu wird ein Bandwidth Broker benötigt. Die Benutzung von RSVP wird diskutiert. Insgesamt entsteht so eine leistungsfähige aber auch komplexe Architektur. Es existiert ein Migrationspfad von geringer zu hoher Komplexität.


CF98 [CF98]: "`Explicit Allocation of Best Effort Packet Delivery Service"'
The paper introduces the Expected Capacity Framework which is similar to DiffServ AF. It allows sender based and receiver based (via ECN) control. Possible services are shown. RIO and a TSW (time sliding window) tagger is introduced. Simulations show good performance with Reno and SACK. The reserved rates are reached independent of RTT. For each connection a tagger is used, nonresponsive flows get a reservation of 0. Paper also available at: http://diffserv.lcs.mit.edu/Papers/exp-alloc-ddc-wf.ps


BBS98 [BBS98]: "`Is Service Priority Useful in Networks"'
Simulationsstudie. Sind meherere Prioritätsstufen für Controlled Load Service sinnvoll? Die Adaptionsfähigkeit der Anwendungen wird einbezogen (Ausspielzeitpunkt am Empfänger / Empfangspuffer). Zur Bewertung wird die Utility-Funktion eingeführt, welche absolute Verzögerung und Variation der Verzögerung einbezieht. Es werden nur Echtzeitanwendungen betrachtet. Es wird keine endgültige Aussage getroffen, ob Priorität sinnvoll ist.


KLT98 [KLT98]: "`Evaluation of Bandwidth Assurance Service using RED for Internet Service Differentiation"'
siehe Abschn. 8.2.


XN99 [XN99]: "`Internet QoS: the Big Picture"'
Übersicht über das Thema QoS in IP-Netzen. Es werden IntServ, RSVP, DiffServ MPLS und Routingtechniken beschrieben und im Kontext von QoS verglichen.
siehe auch Abschnitt 7


GMJ99 [GMJ99]: "`Effect of Number of Drop Precedences in Assured Forwarding"'
Es wird untersucht, ob TCP von 3 drop levels bei AF profitieren kann. Simulationen zeigen, dass 2 levels ausreichen. available at http://www.cis.ohio-state.edu/~jain/papers/dpstdy_globecom99.htm


STK99 [STK99]: "`A Quantitative Study of Differentiated Services for the Internet"'
Es wird das Verhalten bzgl. Verluste und Verzögerung untersucht. Dabei werden analytische Verfahren (Markov-Kette, Fluid Flow)und Simulation eingesetzt. Es wird Threshold Dropping, Priority Scheduling Edge-Discarding und Edge-Marking untersucht. Es kommen Poisson-Verkehr, On-Off-Verkehr und TCP-gesteuerter Verkehr zum Einsatz.


GM99 [GM99]: "`Transport of IP Controlled-Load Service over ATM Networks"'
Simulationsstudie. Vergleicht die Eignung von CBR und nrt-VBR für den Transport von Controlled Load Service (IntServ). Untersuchung sowohl auf Call als auch auf IP-Ebene. Untersuchungsschwerpukt ist der IP/ATM-Router (Puffer, Input Buffer, Token Buffer). CBR und VBR sind fast gleichgut geeignet, Dimensionierungen für den Router werden abgeleitet. Einfache Modelle kommen zum Einsatz (kein Heavy-Tail Filegrößen, FIFO).


CP99 [CP99]: "`An Architecture for Noncooperative QoS Provision in Many-Switch Systems"'
Es wird eine Architektur (SBS=Stratified Best Effort) vorgestellt, das adaptiv die Zugehörigkeit zu einer DiffServ-Klasse für jeden Router festlegt. Dazu wird Lagransche Optimierung benutzt. Die Einstellung am Netzzugang wird per Feedback vorgenommen. Im Netz müssen keine Verbindungsdaten gespeichert werden. Die Simulation zeigen die Vorteile gegenüber fester Reservierung (Verschwendung von Res.) und normalem Best-Effort.


IDBern99 [IDBern99]: "`Interoperation of RSVP/Intserv and Diffserv Networks"'
Internet-Draft. Es wird ein Framework vorgestellt, welches die Zusammenarbeit von IntServ und DiffServ ermöglicht. Dabei wird IntServ am NW-Rand angeboten, DiffServ im Kernnetz. Dabei werden die Fälle unterschieden, ob das Kernnetz (DiffServ) RSVP-Informationen auswerten kann oder nicht.


IDYPG99 [IDYPG99]: "`A Framework for Policy-based Admission Control"'
Internet-Draft. IntServ und RSVP haben das Problem, daß bei Admission Control entschieden werden muß, ob ein Ruf angenommen werden kann. Neben den verfügbaren Ressourcen muß z.B. der Nutzer auch berechtigt sein, die Ressourcen zu belegen. Es wird ein Client/Server-basiertes Framework vorgestellt, welches das ermöglicht. Dieses kann mit IntServ/RSVP zusammenarbeiten.


IDYHBB99 [IDYHBB99]: "`SBM (Subnet Bandwidth Manager): A protocol for RSVP-based Admission Control over IEEE 802-style networks"'
Internet-Draft. Es wird ein Signalisierungsprotokoll definiert, welches de Zusammenarbeit von RSVP (intserv) und Ethernet ermöglicht. Es wird das genaue Verhalten und die Zusammenarbeit mit RSVP spezifiziert. Auf Probleme, wie die nicht durchsetzbaren Garantien in geteilten Medien wird nur hingewiesen.


SNP99a [SNP99a]: "`Study of TCP and UDP Interaction for th AF PHB"'
Investigates performance of AF DiffServ by measurements. The scenario uses 4 AF streams with 6 TCPs and 2 UDP streams with 1 Mbit/s. The bottleneck is 5 Mbit/s and uses RED and RED-3 (called MRED there). Different combinations of marking for TCP (IN/OUT) and UDP (IN/OUT) are considered. In the experiments no fairness between TCP and UDP could be achieved.


BC00 [BC00]: "`A rate adaptive shaper for differentiated services"'
see 8.2


FKSS99 [FKSS99]: "`Understanding and Improving TCP Performance Over Networks with Minimum Rate Guarantees"'
If TCP is used with a token bucket (for AF), tokens are lost because of the TCP congestion control. Guarantees cannot be used. Increasing the bucket size (MBS) can only solve the problem, if the bucket size is very large. In the case of congestion for ACKs a large MBS cannot improve performance. Furthermore, large bursts of IN packet are a problem itself, because they can lead to losses of IN packets or buffer must be dimensioned very large. The paper investigates 3 possible changes to TCP:




FKSS99b [FKSS99b]: "`Adaptive Packet Marking for Maintaining Ent-to-End Throughput in a Differentiated-Services Internet"'
Proposes and evaluates adaptive packet marking. 2 principal mechanisms are investigated. With mechanism 1, Packet are marked with a certain probability p as IN. p is controlled by the actual rate. If rate < target, p is increased, otherwise decreased. It is necessary to limit the changes, otherwise large bursts of IN packets are produced. This is done by an algorithm similar to TCP. Another mechanism is integrated in the TCP source. The results show improved performance, however no means to control the rate of IN packets (leaky bucket, or token, window mechanism etc.) is used in this investigation. With nonresponsive UDP traffic, the approach does not work. in this case, FRED is used to protect the TCP flows.


YR99 [YR99]: "`Modeling TCP Behavior in a Differentiated Services network"'
Es werden einfache Modelle vorgeschlagen, um TCP (Reno) in einer DiffServ-Umgebung zu modellieren. Es werden Modelle für 2 und 3 drop precedences entwickelt (für über und unterlastetes NW). Es wird auch ein TCP mit 2 CWNDs (reserviert + excess) modelliert. Die Modelle zeigen gute Übereinstimmung mit Simulationen. Es wird sich nur auf TCP beschränkt. Es kommt RED zum Einsatz.


DS99 [DS99]: "`Relative Differentiated Services in the Internet: Issues and Mechanisms"'
see 8.1


ECP99 [ECP99]: "`Usefulness of three drop precedences in Assured Forwarding service"'
Internet-Draft. Investigates how 3 drop levels are useful for AF compared to 2 levels. Simulations with TCP SACK in 3 LANs with different CIR are performed. In LAN 4 a UDP source sends with a rate which is varied from 0 to 60 Mbit/s (bottleneck is 34 Mbit/s). For 3 drop precedences, UDP traffic above CIR gets drop precedence 3, TCP gets 2. In the case of 2 precedences, all not conforming traffic is level 2. The results show, that 3 drop precedence levels are useful, because TCP and UDP can be isolated. In case of low reservation level (CIR small compared to bottleneck) CIR can be guaranteed in both cases. For high reservation level guarantees can be met only with 3 levels. If a rate adaptive shaper is used with 2 levels, this can improve performance but still guarantees cannot always be met. The excess is shared very unfair: with 2 drop levels unfair to the TCP, with 3 drop levels, UDP cannot use any excess bandwidth. CBS has no big influence.


DGP99 [DGP99]: "`Quality of Service in the Internet"'
An easy to understand introduction to the topic QoS in the Internet.


SNP99 [SNP99]: "`Bandwidth Assurance Issues for TCP flows in a Differentiated Services Network"'
Messtechnische Untersuchung der AF-Klasse eines DiffServ Netzes. Es wird untersucht, ob die Garantien bei AF eingehalten werden, wenn Parameter wie RTT, Anzahl Microflows, Target Rate, Packet Size, Vorhandensein von Non-Responsive Flows verändert werden. Die Ergebnisse zeigen, dass die Garantien immer eingehalten werden, aber die zusätzliche Bandbreite stark von den Par. abhängt. Der verwendete UDP traffic hat allerdings nur geringe Bitrate.


Hilg99 [Hilg99]: "`Differentiated Services - Konzept und erste Erfahrungen"'
untersucht messtechnisch DiffServ: RED ist besser als drop tail. Fair Queuing bringt keine Verbesserung (wahrscheinlich durch Verluste)


DR99 [DR99]: "`A Case for Relative Differentiated Services and the Proportional Differentiation Model"'
IntServ und DiffServ werden vorgestellt. Es wird auf Relative und Absolute Service Differentiation eingegaben, wobei nur Rel. SD näher betrachtet wird. Es werden ein ein proportional delay Scheduling-Mechanismus WTP (waiting time priority) und ein proportional loss rate dropper vorgestellt, mit dem sich DiffServ realisieren läßt. siehe auch Abschn. 8.1 und 8.2


TM41DIFF [TM41DIFF]: "`Addendum to TM 4.1: Enhancements to Support IP Differentiated Services and IEEE 802D over ATM Draft"'
Enhancement to Traffic Management Specification to add support for DiffServ and IEEE 802.D


TJL99 [TJL99]: "`Attaining per flow QoS with class-based differentiated services"'
Es wird DiffServ untersucht. Dabei kommt als Scheduling-Algorithmus Extended Virtual Clock zum Einsatz, welcher Servicedifferenzierung bzwl. Verzögerung erreicht. Als Puffermanagment wird RED, bzw. DI-RO (Deterministic In - RED OUT). Es werden 4 AF Klassen untersucht. Simulationen mit TCP zeigen gute Ergebnisse, allerdings existiert Unfairness. Es wird nur TCP betrachtet. available from www-run.montefiore.ulg.ac.be/publications/papers/papers-list.html


KS99 [KS99]: "`Performance Comarison of Different Class-and-Drop Treatment of Data and Acknowledgements in DiffServ IP Networks"'
Investigates the influence of the DiffServ class of TCP Acks by simulations. The DiffServ class of Acks has big influence (EF is best), the drop precedence has little influence. (but in the simulation the drop precedence of different flows is fixed and not influenced by the load, no token bucket marker!) The simulation scenario consists of 100 TCP sources (50 sending in each direction), n-RED or RIO queues, TCP can premark packets, TCPs are infinite sources.


RLMA99 [RLMA99]: "`Matching Differentiated Services PHBs to ATM Service Categories - the AF to VBR Case"'
Investigates DiffServ AF by measurements (Cisco Routers, Linux PCs, ttcp, tcpdump, ping). Considers one ATM VBR.1 VC with a shaper. The shaper performs n-RED. One AF class with: 1) single drop precedence and 2) 3 precedences are considered. The different precedences are given to different TCP flows. A differentiation is possible with appropriate setting of RED. The FIFO scheme leads to long RTTs.


NC99 [NC99]: "`A Versatile RED-based Buffer Management Mechanism for the Efficient Support of Internet Traffic"'
Different GFR implementations are evaluated: FIFO with Double EPD (1) and WFQ with per-VC thresholds (+ HBO and LBO as in Double EPD) (2). 2 Scenarios are investigated: Single Bottleneck and Fairness Configuration. With a single TCP per GFR VC guarantees cannot be met with both (1) and (2) in the bottleneck configuration. With multiple TCPs guarantees can be met. (2) behaves always better than (1) and has higher utilization. Without special treatment GFR.1 receives much more bandwidth than GFR.2 if mixed. A partial buffer sharing (needs a per-connection threshold) solves the problem. For single TCP per GFR connection, a RED enhancement of (2) can guarantee the MCR in all cases. No UDP traffic has been used.


EHMV+00 [EHMV+00]: "`Implementing Integrated and Differentiated Services for the Internet with ATM Networks: A Practical Approach"'
introduces the ELISA framework to combine IntServ (Edge) and DiffServ (Core). The edge device performs necessary adaptions (reservation, scheduling, mapping). The framework is optimized for ATM in the core. Annot.: is free to integrate new scheduling disciplines (e.g. FQSM?).


Cnod00 [Cnod00]: "`Rade Adaptive Shapers for Data Traffic in DiffServ Networks"'
The performance of Rate Adaptive Shapers is investigated for transport of TCP. 2 Shapers are used RA-Shaper and Green Shaper. The simulation scenario consists of 10 AF streams with trTCM using different CIR and PIR. The CBS and PBS are relatively small. Users 1-5 do not use the shapers, the other do. Each AF stream contains n multiplexed TCP (SACK). Both shapers perform well, if the number of TCPs is small. If it is large, the gain is only small. With 1 TCP up to 7 times the throughput could be achieved. The green shaper performs better with a small number of TCPs, the RA-shaper with a larger number.


CWZ00 [CWZ00]: "`Rainbow Fair Queueing: Fair Bandwidth Sharing Without Per-Flow State"'
The paper proposes a scheme similar to CSFQ called Rainbow Fair Queuing (RFQ). At the ingress packets are marked with a certain color c. (In contrast, with CSFQ the packets contain the rate itself.) The core routers drop colors larger than a variable color threshold C. C is controlled depending on the congestion level. The scheme keeps the core routers very easy, because it has only to drop colors and must not compute actual rates. Simulations show good performance comparable to CSFQ. Layered applications (e.g video) can exploit RFQ.


Bla00 [Bla00]: "`Differentiated Services and Tunnels, RFC 2983"'
Shows, which problem can occur if tunnels are used together with DiffServ. Possible tunnels are e.g. IP, IPSec or PPP. The problem is due to the fact, that the DS-field of encapsulated packet is not visible outside. There are different possibilities, e.g. copy the inner DS-field to the outer header, etc, which are throroughly discussed.


AWP00 [AWP00]: "`A Fair Traffic Conditioner for the Assured Service in a Differentiated Services Internet"'
proposes a Fair Traffic Conditioner (FairTC). The aim of this conditioner is to allocate a fair share of the tokens for CIR and PIR rate to each flow. It uses a trace queue and a state table. Furthermore, a Fair Multiple RED (FMRED) is used in the border routers. A scenario with 2 behavior aggregates (BA), each containing 1 UDP and 3 Reno TCP flows has been investigated with ns. FairTC together with FMRED can ensure that each flow uses a fair share of the CIR, also the excess is shared fair. However, the authors admit that the mechanisms are sensitive to parameter settings of FairTC and FMRED.


FKR00 [FKR00]: "`A TCP-friendly traffic marker for IP differentiated services"'
proposes to use a TCP-friendly marker at the boundary router of DiffServ to enhance TCP performance over AF. The marker tries to protect flows with small windows, therefore it has to observe the TCP header. It furthermore tries to maintain optimum spacing between IN and OUT packets of a flow, and it marks packets according to the allocation. Simulations of TCP traffic with a "single bottleneck scenario" (no unresponsive traffic) have been performed. They show less timeouts and low packet losses if the TCP-friendly marker is used. Also the avg. TCP goodput increases, but not in all scenarios.


Mue00a [Mue00a]: "`Guaranteed QoS for TCP flows in ATM-based DiffServ Networks"'
Introduces FQSM, a scheme to guarantee per-flow TCP goodput in ATM networks employing GFR.2 or VBR.3 in the presence of unresponsive traffic with high bitrate.


Mue00b [Mue00b]: "`QoS Guarantees for TCP Flows in Differentiated-Services Networks"'
Investigates the performance of Fair Queueing and Selective Marking (FQSM) which has been originally developed for ATM systems. FQSM combines scheduling and marking to obtain guarantees for TCP goodput in the presence of unresponsive traffic with high arrival bit rate. IEEE catalogue number 00Ex423.


Rie00 [Rie00]: "`Simulative Untersuchung eines Verfahrens zur Dienstgütegarantie in DiffServ IP Netzen"'
Investigates TCP performance using per-flow leaky buckets, IN-fair marking, BW-fair marking, (both adaptive and non-adaptive), and FQSM in the presence of unresponsive bitrate with high arrival bitrate. For traffic conditioning of the behaviour aggregate TSW is used. If CBS is large all methods work well. For low CBS methods FQSM provides strong improvements compared to the other methods.


MPLS




BSJ99 [BSJ99]: "`Quality of Service using Traffic Engineering over MPLS: An Analysis"'
Einfache Simulationsstudie. TCP und UDP Verkehrsströme werden transportiert. Werden TCP und UDP in verschiedenen Trunks transportiert, und erfolgt CBQ-Scheduling, ergibt sich gute Performance, sonst unfaires Verhalten.
siehe auch Abschnitt 4


XN99 [XN99]: "`Internet QoS: the Big Picture"'
Übersicht über das Thema QoS in IP-Netzen. Es werden IntServ, RSVP, DiffServ MPLS und Routingtechniken beschrieben und im Kontext von QoS verglichen. siehe auch Abschnitt 6


Li99 [Li99]: "`MPLS and the Evolving Internet Architexture"'
beschreibt MPLS und die Auswirkungen auf die Internet-Architektur. MPLS hat Vorteile fuer Traffic Engineering. Es erfordert Änderungen der Forwarding Architecture. Es werden cie verbindungsorientierte Architektur, die Rolle von MPLS in der Internet-Architektur, Virtual Circuit Emulation, Circuit Protection und Hierarchical Forwarding beschrieben.


Awdu99 [Awdu99]: "`MPLS and Traffic Engineering in IP Networks"'
beschreibt, wie MPLS zu Traffic Engineering eingesetzt werden kann. Es erfolgt ein Vergleich mit dem ATM Overlay Modell. Es werden diskutiert: Path management, Traffic Assignment, Network State Information Dissemination und Network Management.


GJFA+99 [GJFA+99]: "`Traffic Engineering Standards in IP Networks Using MPLS"'
stellt MPLS unter dem Aspekt Traffic Engineering vor. Dabei wird vor allem auf die Signalisierprotokolle RSVP (und notwendige Erweiterungen) und CR-LDP (Constrained Based Routing - Label Distribution Protocol) eingegangen.


CO99 [CO99]: "`Reliable Services in MPLS"'
MPLS wird erklärt. Das Packet Forwarding, QoS Routing and LDP wird vorgestellt. Dann wird auf die Frage Reliability bei Fehlern eingegangen.


Swal99 [Swal99]: "`MPLS Advantages for Traffic Engineering"'
Die Architektur von MPLS wird vorgestellt. Es wird besonders auf Aspekte des Traffic Engineering eingegangen wie LSP tunnels, Constrained Based Routing, Rerouting, ...


Armi00 [Armi00]: "`MPLS: The Magic Behind the Myths"'
discusses MPLS vs. conventional IP forwarding and routing. MPLS makes traffic engineering flexible and easy. However, conventional gigabit routers allow similar performance today. Queuing and scheduling methods may be applied to both.


RVC01 [RVC01]: "`Multiprotocol Label Switching Architecture"'
RFC 3031, This document specifies the architecture for Multiprotocol Label Switching (MPLS).

Scheduling / Buffer Management


Scheduling




DKS90 [DKS90]: "`Analysis and Simulation of a Fair Queueing Algorithm"'
n.v. Also in Proc. ACM SIGCOMM '89, pp 3-12. Defines WFQ


Par91 [Par91]: "`A Generalized Processor Sharing Approach to Flow Control in Integrated Services Networks"'
o. a. PhD Thesis. Definiert GPS und Packet by Packet GPS (das ist WFQ)


Zha91 [Zha91]: "`Virtual Clock: A New traffic control Algorithm for packet switching networks"'
n.a. definiert den Scheduling-Algorithmus Virtual Clock.


KSC91 [KSC91]: "`Weighted round-robin cell multiplexing in a general-purpose ATM switch chip"'
introduces WRR
The authors present the architecture of a general-purpose broadband-ISDN (B-ISDN) switch chip and, in particular, its novel feature: the weighted round-robin cell (packet) multiplexing algorithm and its implementation in hardware. The flow control and buffer management strategies that allow the chip to operate at top performance under congestion are given, and the reason why this multiplexing scheme should be used under those circumstances is explained. The chip architecture and how the key choices were made are discussed. The statistical performance of the switch is analyzed. The critical parts of the chip have been laid out and simulated, thus proving the feasibility of the architecture. Chip sizes of four to ten links with link throughput of 0.5 to 1 Gb/s and with about 1000 virtual circuits per switch have been realized. The results of simulations of the chip are presented.


PG92 [PG92]: "`A generalized processor sharing approach to flow control in integrated services networks -- the single node case"'
n.a. definiert GPS und packet by packet GPS (das ist WFQ)


Gol94 [Gol94]: "`A Self-Clocked Fair Queueing Scheme for Broadband Applications"'
n.a. Stellt SCFQ vor.


SV95 [SV95]: "`Efficient Fair Queuing using Deficit Round Robin"'
also as: Washington University Computer Science Technical Report WUCS-94-17. Describes Defizit Round Robing, a new approximation of fair queuing. Our scheme achieves nearly perfect fairness in terms of throughput, requires only O(1) work to process a packet, and is simple enough to implement in hardware. Deficit RR is also applicable to other scheduling problems where servicing cannot be broken up into smaller units, and to distributed queues.


Zha95 [Zha95]: "`Service disciplines for guaranteed performance service in packet swichting networks"'
n.a. Überblick über Scheduling-Strategien.


Ros96 [Ros96]: "`Resource Allocation on the Cell Level: Link Scheduling and Buffer Management"'
Einfach verständliche Übersicht über Link Scheduling Mechanismen.


Sti96 [Sti96]: "`Traffic Scheduling in Packet-Switched Networks: Analysis, Design, and Implementation"'
PhD Thesis. Einleitung erläutert Scheduling Mechanismen. Ein Algorithmus für Bandbreitenallokierung in input-buffered switches wird vorgeschlagen. Das Modell Latency Rate server wird entwickelt. Eine neue Klassen von Schedulern: Rate-Proportional Server (RPS) wird vorgeschlagen. Diese weisen etwa gleiche Eigensch. wie WFQ auf. Es lassen sich Scheduler designen, die mit geringerer Komplexität ähnliche Leistung wie WFQ erreichen.


Jain97 [Jain97]: "`A Survey of Scheduling Methods"'
Slides; Es werden verschiedene Schedulingverfahren vorgestellt, klassifiziert und kurz bewertet (FIFO, RR, Priority Queueing, PQ with Windows, GPS, Virtual Clock, WFQ, WF2Q, WF2Q+, SCFQ, Stop and Go, Rate Controlled Service Discipline). Referenzliste ist im Internet zu finden.


Rob97 [Rob97]: "`Virtual Spacing for Flexible Traffic Control"'
Virtual Spacing, eine Variante von WFQ wird vorgestellt. Dieses ist identisch mit Self Clocked Fair Queueing. Zunächst wird die Notwendigkeit von WFQ-Verfahren (bzw. Nachteil von FIFO) herausgetellt. Es wird Virtual Spacing und Anwendungen (Video connections, Telefon Service, VPN, ...) erklärt. Es werden Ausführungen zur Realisierung gemacht.


VS97 [VS97]: "`Hardware Implementation of Fair Queueing Algorithms for Asynchronous Transfer Mode Networks"'
Presents approaches to implement scheduling alg. in hardware. They review scheduliers like GPS, WFQ, Virtual Clock. It begins by presenting a general methodology for the design of timestamp-based fair queueing algorithms that provide the same bounds on end-to-end delay and fairness as those of Weighted Fair-Queueing. Based on this general methodology two specific algorithms are described. Illustrative implementations in hardware are discussed.


SSZ98 [SSZ98]: "`Core-Stateless Fair Queueing: Achieving Approximately Fair Bandwidth Allocations in High Speed Networks"'
Ein Verfahren zur fairen Verteilung von Bandbreite wird vorgestellt (CSFQ). Es basiert darauf, am NW-Rand die Bandbreite eines Flows festzustellen und dann im Core darauf basierend Pakete zu verwerfen. Im Kernnetz wird keine Speicherung von Flow-Daten benötigt. Das Verfahren wird analysiert und gegen andere Verfahren (FIFO, RED, FRED, DRR) verglichen.


CGE98 [CGE98]: "`Time-Shift Scheduling - Fair Scheduling of Flows in High-Speed Networks"'
Ein Scheduling Protocol: Time-Shift Scheduling (TSS) wird vorgestellt. Zunächst werden die angestrebten Eigenschaften eines Schedulers genannt. Es wird ein Überblick über Scheduling-Protokolle gegeben. Insbesondere Virtual Clock, Weighted Fair Queueing und Self-Clocking Fair Queueing werden näher betrachtet. Es wird nun TSS vorgestellt und dessen Eigenschaften analysiert. TSS erfüllt alle zu Beginn genannten Anforderungen (Rate-proportional delay, Efficiency, Fairness).


SV98 [SV98]: "`Rate-Proportional Servers: A Design Methodology for Fair Queueing Algorithms"'
Es wird beschrieben, wie beim Design von Fair Queueing (FQ) Algorithmen vorzugehen ist. Es wird eine Einführung in FQ gegeben (incl. Literaturverweise). Das Problem ist die hohe Komplexität von O(N) beim Emulieren von Generalized Processor Sharing . Es wird die Oberklasse der Rate-Proportional Server eingeführt, der auch z.B. WFQ angehört. Es werden die Eigenschaften abgeleitet. Basierend auf diesem Framework werden in einer Referenz zwei FQ-Algorithmen entwickelt, welche nur eine Komplexität von O(1) für das Erzeugen des Timestamps aufweisen.


BLN99 [BLN99]: "`Fair Queuing in Wireless Networks: Issues and Approaches"'
Es werden die Probleme des Fair Queuings in drahtlosen Netzen diskutiert. Das Problem ist, daß durch Fehler auf dem Link kurzzeitig nicht genutzte Bandbreite später genutzt werden soll. Im Festnetz dagegeben ist nicht genutzte andbreite verloren. 4 Ansätze (Idealized WFQ, Channel-Condition Independent FQ, Server Based Fairness Approach und Wireless Fair Service werden gegenübergestellt.


NSSZ99 [NSSZ99]: "`Supporting Best-Effort Traffic With Fair Service Curve"'
Bei Packet Fair Queueing ist Verzögerung und Durchsatz eng gekoppelt. Das führt zu relativ hoher Verzögerung für kurze Flows. Es wird vorgeschlagen, kurze Flows mit höherer Rate zu bedienen. Dadurch haben kurze Flows eine geringe Verzögerung, lange Flows werden nicht behindert. Es werden Simulationsergebnisse vorgestellt, welche den Vorteil des Verfahrens zeigen.


DS99 [DS99]: "`Relative Differentiated Services in the Internet: Issues and Mechanisms"'
Es werden Scheduling-Mechanismen untersucht, inwieweit sich DiffServ bzgl. Verzögerung realisieren läßt. Untersucht werden: Strict Priorities, WFQ, Earliest Deadline First (EDF) und Waiting-Time Priority (WTP). WPT eignet sich am besten, da hier Verzögerung und Rate entkoppelt sind. EDF eignet sich nur bei hoher Last. siehe auch Abschn. 6


SZ99 [SZ99]: "`Providing Guaranteed Services Without Per Flow Management"'
Es wird vorgeschlagen, mittels eines verteilten Algorithmus das Scheduling Verfahren Jitter Virtual Clock, welches eine Art des Scheduling-Verfahrens Virtual Clock (VC)ist, nachzubilden. Dieses Verfahren wird als Core-Jitter-VC bezeichnet. Es benötigt keine Flow-spezifische Informationsspeicherung. Informationen werden im IP-Header übertragen. Um QoS zu realisieren, wird CAC durchgeführt. vailable at http://www.cs.berkeley.edu/~istoica/QoS/


KR99 [KR99]: "`Comparison of Service Disciplines in ATM Networks Using a Unified QoS Metric"'
Es werden die Schedulingverfahren FCFS, WFQ, Strict Priority, Delay-Earlies Due Date und Jitter-Earliest Due Date für ATM-Netzwerke untersucht. Zum Vergleich wird für jeden QoS-Größe (Loss, Delay, Jitter) das Produkt aus Leistungskenngrößen der Form 1/(1+x/e) verwendet (x: wert, e: expected). Es wurde der Simulator YATS verwendet.


MBLM99 [MBLM99]: "`A QoS-Aware Scheduling Algorithm for Input Queued Switches"'
Es wird ein Scheduling-Verfahren (Fair Reservation with Preemption and Acknowledgement, F-RPA) für eingangsgepufferte Switche vorgeschlagen und untersucht. F-RPA ist einfach und flexibel. Es erlaubt verschiedene Verkehrsklassen und Bandbreitengarantie. Die Leistungsfähigkeit wird durch Simulationen gezeigt. Das Verhalten bzgl. Verzögerung ist verbesserungswürdig.


DR99 [DR99]: "`A Case for Relative Differentiated Services and the Proportional Differentiation Model"'
IntServ und DiffServ werden vorgestellt. Es wird auf Relative und Absolute Service Differentiation eingegaben, wobei nur Rel. SD näher betrachtet wird. Es werden ein ein proportional delay Scheduling-Mechanismus WTP (waiting time priority) und ein proportional loss rate dropper vorgestellt, mit dem sich DiffServ realisieren läßt. siehe auch Abschn. 6 und 8.2


CG00 [CG00]: "`Implementation and Experimental Analysis of IP Schedulers in an IntServ-over-ATM Test-bed"'
Propoerties of schedulers are listed. Schedulers are compared: WRR, WFQ, Virtual Clock (VC), Worst-case Fair Weighted Fair Queueing (WF2Q), Shaped Starting Potential Fair Queueing (S-SPFQ). S-SPFQ has been investigated by measurements in a testbed with respect to the following criteria: service rate limitation, spare bandwidth redistribution,flow isolation, fairness.


Buffer Management




FJ93 [FJ93]: "`Random Early Detection Gateways for Congestion Avoidance"'
RED wird vorgestellt. Es wird die Verlustwahrscheinlichkeit in Abhängigkeit der Queuelänge gesetzt. Simulationen zeigen die Vorteile (keine Benachteilung von Bursthaftem Verkehr, keine globale Synchronisation). Die Erkennung von misbehaving users wird vorgeschlagen.


Fore96 [Fore96]: "`ForeThought Bandwidth Management, version 1.0"'
beschreibt das Traffic Managment des Fore ASX-200. Es kommen per flow scheduling, smart buffers, PPD, EPD zum Einsatz siehe auch Abschn. 9


GH96 [GH96]: "`UBR+ Service Category Definition"'
benutzt und definiert Double-EPD
siehe auch Abschn. 9


Turn96 [Turn96]: "`Maintaining high throughput during overload in ATM switches"'
untersucht EPD, schlägt EPD mit Hysterese bzw. Fair EPD mit Hysterese vor. Dadurch wird der Pufferbedarf stark verringert. siehe auch Abschn. 9


LM97 [LM97]: "`Dynamics of Random Early Detection"'
Es wird die Effektivität von RED untersucht. Die Nachteile werden durch Simulationen aufgezeigt. FRED wird vorgeschlagen. Dabei wird die Verlustwahrscheinlichkeit durch die Anzahl der Pakete des Flows in der Warteschlange bestimmt. Die Vorteile werden gezeigt. Es werden Erweiterungen für sehr viele aktive Flüsse vorgeschlagen.


FKSS97 [FKSS97]: "`Techniques for Eliminating Packet Loss in Congested TCP/IP Networks"'
Simulationsstudie. RED hat durch den festen Parameter $p_{max}$ Probleme bei wechselnder Anzahl von Verbindungen. $p_{max}$ wird mit Hilfe der Queuelänge dynamisch angepasst. Es wird gezeigt, das ECN (Early Congestion Notification) notwendig ist und Packet Drop problematisch. Eine Erweiterung für viele Verbindungen (SUBTCP) und damit lange Intervalle zwischen Paketen wird vorgeschlagen.


EA97 [EA97]: "`RED Algorithm in ATM Networks"'
Eine Erweiterung von RED für ATM-Netzwerke (C-RED) wird vorgeschlagen. Es wird TCP-Verkehr untersucht. Als Hingergrund wird On-Off Verkehr verwendet. Es laesst sich vergleichbare Performance wie mit RED erzielen.


BCCD+98 [BCCD+98]: "`Recommendations on Queue Management and Congestion Avoidance in the Internet"'
RFC 2309, points out the need to implement active queue management. RED is recommended as the standard buffer management algorithm. It provides reduced packet loss, lower delay, avoid lock-out behavior. The document deals with unresponsive and non TCP compatible protocols. As the amount of such traffic is expected to increase, the document points out, that this problem has to be addressed by future research. see also section 4.


Amit98 [Amit98]: "`Routers Mechanisms for Congestion Avoidance"'
Eine Verbesserung fuer RED wird untersucht. Es werden die TCP-Flows mit vielen Verlusten innerhalb eines Intervals in einer niederpriorisierten Queue bedient. Es werden kurze und lange Flows betrachtet.


KLT98 [KLT98]: "`Evaluation of Bandwidth Assurance Service using RED for Internet Service Differentiation"'
Simulationsstudie. Für ein erweitertes RED (RED-x2) wird untersucht, inwieweit sich Garantien für DiffServ einhalten lassen. Für RED werden 2 Prioritäten eingeführt, die unabhängig max_th, min_th, max_p und q_avg für in und out-Pakete berechnen. Verändert wird die Bandbreite der Profile (wobei 2 Gruppen verwendet werden). Es wird TCP (infinite source und WWW) und UDP-Verkehr verwendet, 2 Szenarien: Mux und Parking Lot, verschiedene RTTs. Dabei wird jeweils nur untersucht, wie 1 Verb. (bei infinite) und das Aggregat (bei WWW) die Bandbreite entsprechend des Profils nutzen kann. Es wird die Schlußfolgerung gezogen, daß RED-x2 ausreichend ist, um Garantien geben zu können. siehe auch Abschn 6


KLS98 [KLS98]: "`Beyond Best Effort: Router Architectures for the Differentiated Services of Tomorrow's Internet"'
stellt Router-Architektur vor. Geht auf Fair Queueing ein. Das ist schwierig, da sehr viele (TCP) Flows (256000 bei OC-3) auftreten können. Deshalb wird FQ für Klassen durchgeführt. Router müssen TCP unterstützen (Bursthaft, Unfair, Synchronization, Random Loss Sensitivity). Per Flow Queuing bzw. Puffer Management allein sind nicht ausreichend. Eine Kombination FQ-ALQD (approximated longest queue drop) wird gegen FQ-RED und FCFS-RED verglichen und erreicht wesentlich bessere Ergebnisse bei extensivem CBR-Verkehr (UDP).
siehe auch Abschn. 4.


BC00 [BC00]: "`A rate adaptive shaper for differentiated services"'
Introduces and evaluates a rate adaptive shaper. The shaping rate is determined by the buffer occupancy which smoothes bursts. Simulations show that the shaper can improve TCP performance in a N-source scenarios, see also 6


RBL99 [RBL99]: "`A RED discard strategy for ATM networks and its performance evaluation with TCP/IP traffic"'
Eine Variante von RED für ATM wird vorgeschlagen und für verschiedene Szenarien untersucht. Es erfolgt ein Vergleich gegen plain UBR (drop tail) EPD und FBA. FBA und RED sind fast immer besser geeignet als EPD bzw. drop tail.


DR99 [DR99]: "`A Case for Relative Differentiated Services and the Proportional Differentiation Model"'
IntServ und DiffServ werden vorgestellt. Es wird auf Relative und Absolute Service Differentiation eingegaben, wobei nur Rel. SD näher betrachtet wird. Es werden ein ein proportional delay Scheduling-Mechanismus WTP (waiting time priority) und ein proportional loss rate dropper vorgestellt, mit dem sich DiffServ realisieren läßt. siehe auch Abschn. 8.1 und 6


LJ99 [LJ99]: "`Improving Explicit Congestion Notification with the Mark-Front Strategy"'
Es wird Explicit Congestion Notification untersucht. Dabei wird bei Mark-Front eine bessere Performance als bei Mark-tail (utilization, fairness). Der Grund ist, dass TCP Fast Retransmit schneller ueber verlorene Daten informiert wird. Dadurch wird weniger Puffer benötigt. Es wird RED benutzt. Auch neue Verbindungen profitieren von Mark-Front. siehe auch Abschn. 4


SLSC99 [SLSC99]: "`Buffer Management Schemes for Supporting TCP in Gigabit Routers with Per-Flow Queueing"'
es wird untersucht, welche Auswirkung Scheduling bzw. Buffer Management auf TCP-Flüsse hat. Es werden FCFS und FQ bzw. Tail Drop, RED, LQD (Longest Queue Drop), RND (Auswahl zufälliger Backlogged Flüsse mit mehr als fair Share) und AQLD (Approx. LQD) betrachtet (LQD, ALQD und RND sind Drop from Front). Scheduling allein reicht nicht aus, um Fairness (bzw. Garantien) zu erreichen. Tail Drop hat die schlechtesten Eigenschaften, Drop from Front Strategien sind prinzipiell günstiger, da Fast Retransmit schnell informiert wird. Auch RED ist ungünstig, da kein Wissen über Flows existiert. Es treten Mehrfachverluste auf und wiederholte Pakete werden verworfen. Unter den anderen Verfahren ist FQ-LQD vorzuziehen. Siehe auch Abschn. 4.


LB99 [LB99]: "`Packet Dropping Policies for ATM and IP Networks"'
Packet Dropping methods for IP and ATM networks and its aims (network utilization, application throughput, fairness, simplicity, avoid. global synchronization) are discussed. Mech ATM: Drop Tail, PPD, EPD, Selective Drop, EPD with Fair Buffer Allocation (EPDFBA), Early Selective Packet Discard (ESPD), EPD with Hysteresis, Drop from Front, Age Priority PD (APPD), Preemptive PPD. Mech. IP: RED, FRED, RIO


CEP00 [CEP00]: "`RED Behavior with Different Packet Sizes"'
It is shown, that RED is unfair with respect to MTU size. Version 1, where drop probability is independent of the packet size discriminates connections with small MTUs. Version 2, which takes into account the packet size discriminates connections with high MTU. 3 other RED variants are investigated, which show to improve the behavior. TCP SACK has been used, no UDP traffic, a bottleneck scenario with 20*3 TCPs with 3 differnt MTUs at the same time.


PPP00 [PPP00]: "`Choke, A stateless active queue management scheme for approximating fair bandwidth allocation"'
Unresponsive UDP traffic can cause significant unfairness in IP networks. A buffer management technique called CHOKe is proposed which overcomes this problem. On arrival of a packet, CHOKe compares the packet with a randomly chosen packet (from the queue). If it is from the same flow, both are discarded. With one unresponsive flow, the scheme shows good performance, for more it fails. If there are k UDP flows k comparisons should be made. The scheme must then be adjusted to the correct value of k. The authors propose to adjust this value with respect to queue length.


ATM




I.321 [I.321]: "`B-ISDN Protocol reference model and its application"'
Recommendation I.320 contains the description of the ISDN protocol reference model (PRM).


UNI3.0 [UNI3.0]: "`ATM User-Network Interface Specification, Version 3.0"'
specifies the ATM User Network Interface (UNI)


ATMFTM40 [ATMFTM40]: "`Traffic Management Specification Version 4.0"'
Grundlegendes Standarddokument. Beschreibt Serviceklassen, Quality of Service für ATM, Traffic Contract und Funktionen für Traffic Management.


ATMFTM41 [ATMFTM41]: "`Traffic Management Specification Version 4.1"'
Grundlegendes Standarddokument. Beschreibt Serviceklassen, Quality of Service für ATM, Traffic Contract und Funktionen für Traffic Management.


Laub94 [Laub94]: "`Classical IP and ARP over ATM"'
RFC 1577, defines CLIP, obsoleted by RFC 2225


LH98 [LH98]: "`Classical IP and ARP over ATM"'
RFC 2225, defines CLIP and ARP over ATM, makes RFC 1626 and RFC 1577 [Laub94] obsolete




GH96 [GH96]: "`UBR+ Service Category Definition"'
ATMF Dokument 96-1598, definiert UBR+, eine Erweiterung zu UBR, die Mindestgarantien auf Frameebene gibt, aber keine Signalisierung benötigt. Es werden 2 Beispielimplementierungen vorgestellt: 1. WFQ+per VC queuing; 2. Tagging und FIFO queue (Double-EPD); beide Varianten beachten das CLP-Bit.
siehe auch Abschn. 8.2


Fore96 [Fore96]: "`ForeThought Bandwidth Management, version 1.0"'
beschreibt das Traffic Managment des Fore ASX-200. Es kommen per flow scheduling, smart buffers, PPD, EPD zum Einsatz siehe auch Abschn. 8.2


Turn96 [Turn96]: "`Maintaining high throughput during overload in ATM switches"'
untersucht EPD, schlägt EPD mit Hysterese bzw. Fair EPD mit Hysterese vor. Dadurch wird der Pufferbedarf stark verringert. siehe auch Abschn. 8.2


CLWS96 [CLWS96]: "`Monitoring and Control of ATM Networks Using Special Cells"'
Es wird das Konzept der OAM-Zellen bei ATM vorgestellt. Es wird auf Operation und Management und auf Netzwerk-Management eingegangen.


CIFTut97 [CIFTut97]: "`Cells-IN-FRAMES: THE REALIZATION OF ATM-TO-THE-DESKTOP"'
Erklärt das Verfahren Cell-In-Frames (CIF), um ATM bis zum LAN-basierten Endteilnehmer (z.B. basierend auf Ethernet) zu transportieren.


RobertsCIF97 [RobertsCIF97]: "`Cells In Frames - ATM with Variable Length Packets"'
Das Verfahren CIF (Cells in Frames) wird erklärt. Die Vorteile gegenüber anderen Lösungen (ATM, IP) werden herausgestellt. Eine 100fache Geschwindigkeitssteigerung gegenüber dem heutigen Internet wird in Aussicht gestellt.


BP97 [BP97]: "`GFR Implementation Alternatives with Fair Buffer Allocation Schemes"'
Shows that FIFO cannot provide GFR guarantees even with fair buffer allocation (per-VC accounting). Use TCP and unresponsive UDP traffic with high bitrate. One TCP per VC.


GJF+97 [GJF+97]: "`Simulation Experiments with Guaranteed Frame Rate for TCP/IP Traffic"'
Investigates tagging, per-VC allocation (fair buffer allocation) and per-VC scheduling. Per-VC scheduling is sufficient to provide guarantees to CLP0+1 stream (no network tagging). In the case of network tagging, per VC-scheduling and per-VC buffer allocation are required to give quarantees. TCPs (SACK, one per VC) are used with equal and unequal allocation. No unresponsive UDP traffic is included. The utilization is 100%.


SBGF97 [SBGF97]: "`IP and ATM - a position paper"'
gives an overview on IP technology (e.g. IPv4, IPv6,RSVP), ATM technology and interworking (CLIP, LANE MPOA).


GJFV98 [GJFV98]: "`Buffer Management for the GFR Service"'
Present DFBA (Dynamic Fair Buffer Allocation) and show by simulations with TCP traffic (20 TCPs in one ATM VC, 5 ATM VCs), that this is enough to provide MCR guarantees, especially if the reservation level is low. No unresponsive UDP traffic has been used.


KJFG98 [KJFG98]: "`Performance and Buffering Requirements of Internet Protocols over ATM ABR and UBR Services"'
Es wird das Verhalten von TCP über ABR und UBR untersucht. TCP arbeitet am besten, wenn keine Verluste auftreten. UBR benötigt die Summe der max. Fenster als Puffer im Netzwerk. ABR benötigt wesentlich weniger Puffer. Der Bedarf ist abhängig von RTT. Die Queues in der ABR-Quelle können groß sein. Die Ergebnisse werden simulativ in einer N-Source Konfiguration gewonnen.


Lee98 [Lee98]: "`GFR Dimensioning for TCP Traffic"'
Investigates the dimensioning of the F-GCRA (MBS) for GFR with TCP traffic. Unresponsive traffic is included but is in separate VCs. For dimensioning the MBS parameter, the TCP characteristic has to be taken into account. Especially with coarse timer granularity, the TCP performance may suffer.


HV98 [HV98]: "`Simulation of TCP/IP Router Traffic over ATM using GFR and VBR.3"'
Compares GFR and VBR.3. 5 ATM VCs with different SCR (MCR) transport 10 TCP sources (Reno), there is no UDP traffic. GFR is superior over VBR.3 if the MBS is small (24 cells = 2 Packets). If the MBS is large the difference is small. Guarantees are not met for TCP traffic.


Bona98 [Bona98]: "`Integration of ATM under TCP/IP to provide services with guaranteed minimum bamdwidth"'
Dissertation. Investigates, how TCP can be transported with CBR, VBR and GFR. For GFR simulations are performed and show that the GFR service is able to better support TCP/IP traffic than the VBR service. Several problems with the two example switch implementations are identified and solutions to overcome these problems are given. 1.: shaping algorithm 2.: an alternative switch implementation which is able to completely support the GFR service in simple FIFO-based ATM switches. 3.: modified scheduler which provides a better support for the GFR service than the classical WFQ-like schedulers by decoupling the allocation of the unreserved bandwidth from the provision of the minimum guaranteed bandwidth.
Abstract avail. at http://www.info.fundp.ac.be/~obo/these.html


Goyal99 [Goyal99]: "`Traffic Management for TCP/IP over Asynchronous Transfer Mode (ATM) Networks"'
Dissertation. Umfangreiche Arbeit zum Thema TCP Traffic Management in ATM-Netzen Enthält Überblick über TCP Congestion Control (sehr umfassend) und Puffer-Management (ohne per VC scheduling). Simulativ werden untersucht:

Prinzipiell werden nur einfache Szenarien untersucht. TCP greedy. Langzeit-Durchsatz ist alleiniger Parameter (Effizienz und Fairness). Kurze Simulationen (10s, 20s). Auch wird anderer IP-Verkehr nicht betrachtet. Die Zusammenfassung nennt Verbesserungen von TCP als weitere Aufgabe (TCP friendly buffer management, ECN, ratenbasiert).


Heiss99 [Heiss99]: "`QoS results for the GFR ATM service"'
Zwei Versionen des framebasierten GCRA werden gegenübergestellt. Es werden die Nachteile des CF-GCRA beschrieben, insbesondere die Abhänigkeit der Ergebnisse von der Toleranz. Eine Verbesserung, der F-GCRA wird vorgestellt. Es wird die zu erwartende QoS untersucht. F-GCRA liefert eine bessere QoS-Definition besonders dann, wenn auch nichtkonforme Zellen auf den F-GCRA treffen.


FJRGV99 [FJRGV99]: "`Quality of Service for Internet Traffic over ATM Service Categories"'
Es wird eine Architektur zur Verbindung von VPNs über Internet vorgestellt. Dabei wird ATM als Backbone benutzt. Es werden die ATM service categories verglichen. ABR wird als Favorit herausgestellt.


CG99 [CG99]: "`QoS Control and Interworking of Overbooked Elastic and Brittle ATM Traffic"'
PVC's werden typischerweise überbucht. Es werden die Auswirkungen auf gesteuerten (GV) (TCP) und nichtgesteuerten (NGV) (Video) Verkehr untersucht. Es wird das Verhalten von UBR, VBR.3 und GFR untersucht, in der NGV Priorität bekommt. PPD hat pos. Effekt, die Verkehrsklasse kaum. Die Reservierung einer Mindestrate für GV bringt Vorteile. Es wird vorgeschlagen, CAC getrennt für GV u. NGV vorzunehmen.


Dahl99 [Dahl99]: "`Betrieb und Wartung von ATM-Netzen mit OAM-Zellen"'
OAM wird vorgestellt. Es wird u.a. auf F4, F5 Fluesse, OAM Zelltypen, Fehler eingegangen


CFE99 [CFE99]: "`Performance Evaluation of TCP Extensions on ATM over High Bandwidth Delay Product Networks"'
siehe Abschn. 4.


MHL99 [MHL99]: "`Influence of the ATM Service Category on TCP Performance in Large Network"'
Es wird die Leistungsfähigkeit von TCP in grossen Netzwerken bei vielen Verbindungen untersucht. Es erfolgt ein Vergleich verschiedener ATM Servicekategorie, die Abhängigkeit von Puffergröße und PPD/EPD. PPD hat positiven Effekt. EPD hat nur geringe Auswirkung. UBR kann gut arbeiten, wenn genügend Puffer vorhanden ist. Geringer Puffer hat neg. Auswirkung auf die Fairness und Performance. TCP Applikationen können die Reservierungen von VBR.3 nicht nutzen. Wenn VBR.3 zusammen mit UBR verwendet wird, kann VBR.3 von den Garantien profitieren.


ALPS99 [ALPS99]: "`Providing Rate Guarantees for Internet Application Traffic Across ATM Networks"'
Surveys possibilities to provide rate guarantees with GFR. It introduces into policing, buffer management, and scheduling. Presents and evaluates FIFO and FQ-based (WFQ, WRR) schemes, frame tagging, per-VC queuing, FBA, Weighted Buffer Allocation, DFBA. Even with FIFO based schemes guarantees for goodput of an ATM VC can be given if the right buffer management and policing is used. Different flows inside one ATM connection are not considered.


CH00 [CH00]: "`Balanced Packet Discard for Improving TCP Performance in ATM Networks"'
TCP (except SACK) cannot efficiently recover from multiple losses in one RTT. If more than 3 segments are lost or cwnd < 5 segments this leads always to a timeout. The paper proposes an extension to EPD. Another threshold above the EPD thr. is introduced. After a loss a connection uses this higher threshold for a number of segments (called gap). This reduces the number of timeouts significantly. With SACK the gain is not big, because it can efficiently recover from multiple losses, but it still can improve fairness to other TCP implementations. Problems not covered: multiple TCP flows per ATM connection, multiple hops.


PKD00 [PKD00]: "`A Performance Evaluation of Buffer Management and Scheduling for ATM-GFR Using TCP"'
Overview on TCP/ATM performance problems. Current issues for GFR are given. Some of thhe relevant literature has been summarized. Simulations are performed (with YATS) using a simple scenario and TCP sources (1 TCP per GFR VC). CLR and packet retransmission rate has been measured. As switch the following are used: UBR, EPD, 2 Thresholds, WFQ, per VC accounting.

Switch/Router-Realisierungen




BWBM98 [BWBM98]: "`Traffic Management for an ATM Switch with Per-VC Queueing: Concept and Implementation"'
Es wird die Switch-Architektur von Siemens mit zusteckbaren SMUs vorgestellt, welche im Telecom-Projekt verwendet wurde. Diese unterstützt alle ATM Serviceklassen. Es wird per-VC queueing benutzt. Für non-realtime Klassen wird WFQ angewendet. Traffic Shaping ist möglich. Der Puffer kann pro Verbindungsklasse und Verbindung reserviert werden.


Par98 [Par98]: "`A 50-Gb/s IP Router"'
Es wird eine IP-Routerarchitektur vorgestellt, welche einen aggregierten Durchsatz von 50 Gbit/s erlaubt. Diese besteht aus Line Cards (mit mehreren Interfaces), Forwarding Engines, Switch und Network Processor. Die Beschreibung ist teilweise sehr detailliert. So wird z.B. die Forwarding Engine sehr genau beschrieben, der Prozessorcode wird diskutiert. Gründe für die Auswahl der Architekturmerkmale werden genannt. Der Router erlaubt neben dem reinen Paket-Forwarding auch QoS-Funktionen. Die Anwendung von Queueuing-Strategien wird aber nicht näher beschrieben.
can be found at: www.ir.bbn.com/ craig/router.ps


SV99 [SV99]: "`Fast Address Lookups Using Controlled Prefix Expansion"'
Es werden Methoden für das Suchen von IP Addressen mit longest matching prefix vorgestellt. Es werden verschiedene Methoden von prefix expansion erklärt und untersucht. Anhand von gemessenen Routingtabellen wird die Leistungsfähigkeit untersucht. Die vorgeschlagenen Schemen benötigen nur halb so viele Speicherzugriffe und sind dadurch etwa 2mal so schnell, wie bekannte Algorithmen

Mobile




TTL99 [TTL99]: "`The Challenges of Seamless Handover in Future Mobile Multimedia Networks"'
Es wird Handover in IP-Netzen (bzw. auch ATM) untersucht. Dabei wird zwischen horizontal (gleiches System) vertikal (in ein anderes System, z.B. von DECT nach UMTS) unterschieden. Es wird erklaert, wie Mobile-IP funktioniert und auf das Routing eingegangen. Verbesserungen fuer TCP (I-TCP, Modified I-TCP, Snoop TCP und SACK TCP und ARQ) werden vorgestellt.


VJFD99 [VJFD99]: "`AQuaFWiN: Adaptive QoS Framework for Multimedia in Wireless Networks and"'
Ein Lösung für das Problem QoS in mobilen Netzen wird vorgestellt. Die Applikationen müssen dabei adaptiv sein. Die Netzwerkarchitektur ist hierarchisch aufgebaut. Als MAC wird TDMA gewählt, wobei Slots reserviert werden, verbleibende Bandbreite wird durch CSMA/CD genutzt. QoS-Parameter werden als Bereich angegeben. Es gibt einen Feedback-Mechanismus, der von allen Ebenen ausgewertet wird. Es werden weitere Ansätze vorgestellt (WAMIS, SWAN, MMWM, QGMC) und verglichen.


Rein99 [Rein99]: "`Soft QoS Control in the WATMnet Broadband Wireless System"'
Es wird ein QoS-Konzept für Wireless ATM vorgestellt. Dabei wird Soft-QoS verwendet, d.h. die Nutzerzufriedenheit ist f(Bandbreite), wobei typ. eine S-Form gewählt wird. Neue Stationen können bei Congestion noch angenommen werden (z.B. nach Handover), wenn andere Stationen weniger BB erhalten (entsprechend QoS-Funktion). Die Einbindung in den Protokollstack wird vorgenommen. Simulationsergebnisse werden präsentiert.


LRKOJ99 [LRKOJ99]: "`Multi-Layer Tracing of TCP over a Reliable Wireless Link"'
Es wird ein Tool vorgestellt, welches es erlaubt, Datentransfers mittels TCP/IP in enem GSM-Netz auf verschiedenen Ebenen zu messen und die Ergebnisse zusammenzuführen. Es wurde gezeigt, dass die Performance von TCP meist ideal ist. Probleme treten durch Header Compression und durch starkes Queueing infolge von grossen Puffern auf. Aus diesem Grund sollte in größeren Mobilnetzen aktives Queue-Management eingesetzt werden.

Video Transport




BFJF99 [BFJF99]: "`A Survey of Application Layer Techniques for Adaptive Streaming of Multimedia"'
Es werden Methoden im Endsystem untersucht, um sich an Netzwerkbedingungn anzupassen (smoothing, adaption). Es wird auf Kompressionstechniken (MPEG, Wavelet Encoding) und Ratenadaption (Layered Coding, Receiver Driven Multicast) für Video sowie adaptiver FEC für Audio eingegangen.


Krun99 [Krun99]: "`Bandwidth Allocation Strategies for Transporting Variable-Bit-Rate Video Traffic"'
Es werden verschiedene Ansätze diskutiert, wie die Bandbreite für den Videotransport verringert werden kann. CBR und VBR coding, online/offline, interacivity, adaptiveness. Es werden QoS-Anforderungen erwähnt. Netzmechanismen in ATM und IP (RTP, IntServ, DiffServ) werden betrachtet. Möglichkeiten für Statisches Multiplexen und Multicast werden beleuchtet.

Active Networks




BM98 [BM98]: "`Mobile Agents - Enabling Technology for Active Intelligent Network Implementation"'
Es wird ein Überblick darüber gegeben, wie Mobile Agenten eingesetzt werden können, um IN-Dienste zu realisieren. Anhand der Beispiele Call Forwarding und VPN Service wird die Nützlichkeit deutlich.


Theo98 [Theo98]: "`ACC: Using Active Networking to Enhance Feedback Congestion Control Mechanisms"'
siehe auch TCP


WLG99 [WLG99]: "`Introducing New Internet Services: Why and How"'
Das Konzept des aktiven Netzwerks wird vorgestellt. Dabei sind die Knoten programmierbar. Die Programmierung erfolgt paketweise durch sog. Kapsule. Es werden Beispiele vorgestellt (Auktionssystem, Börsensystem). Die effektive Einführung neuer Dienste wird dadurch erleichtert, das zieht aber eine erhöhte Komplexität der Knoten nach sich und widerspricht damit KISS.

software




GNUTELLA [GNUTELLA]: "`Gnutella"'
Gnutella is a real-time search, peer-based file-sharing client that allows a user running a Gnutella client to search for and download files from other Gnutella users. Available at http://gnutella.wego.com/


ISABEL [ISABEL]: "`ISABEL multiconferencing application"'
software used in ACTS EXPERT project, see http://isabel.dit.upm.es/


MOSEL [MOSEL]: "`MOSEL"'
MOSEL stands for MOdeling Specification & Evaluation Language. It is a computer language to model queueing systems and solve associated markov chains. For solving other software is employed, e.g. Petrinets solver. Software and further information is available at http://www4.informatik.uni-erlangen.de/Projects/MOSEL/


NAPSTER [NAPSTER]: "`Napster"'
a peer-to-peer music sharing application available at http://www.napster.com


ns2 [ns2]: "`ns-2 Simulator"'
A Simulator to simulate IP networks, available at http://www.isi.edu/nsnam/ns/


REALPLAYER [REALPLAYER]: "`RealPlayer"'
a video- and audio streaming application available athttp://www.realplayer.com.


YATS [YATS]: "`YATS - Yet Another Tiny Simulator"'
a simulator to simulate ATM networks


YATSUSER [YATSUSER]: "`YATS - Yet Another Tiny Simulator User's Manual for Version 0.3"'
The YATS users manual.

noch nicht eingeordnete Literatur




BS89 [BS89]: "`Taschenbuch der Mathematik"'
Bronstein; Standardwerk der Mathematik


Jain91 [Jain91]: "`The Art of Computer Systems Performance Analysis: Techniques for Experimental Design, Measurement, Simulation, and Modeling"'
Buch, welches Techniken für experimentelles Design, Messungen, Simulationen und Modellierung beschreibt. Führt ein in die Probleme der Performance Evaluation. Es werden behandelt: Messmethoden, Wahrscheinlichkeitstheorie und Statistik (Meßdaten, Regression), Experimentdesign (Factorial Design), Simulation (RNG, Verification, Validation, Distributions), Queueuing Theorie.


SH95 [SH95]: "`Ereignisorientierte Simulation: Konzepte und Systemrealisierung"'
Small and easy to read book on simulation methodology.


Schwartz96 [Schwartz96]: "`Broadband Integrated Networks"'
Buch. Gibt Überblick über Leistungsbewertung von Datennetzen. Stellt nachvollziehbar Berechnungsmethoden vor. Kapitel: Traffic Characterisation, Admission Control, ATM-Switches, End-to-End traffic bounds and effect. capacity, Feddback Congestion Control, Appendix Elements of Queueing Theory


Kesh97 [Kesh97]: "`An Engineering Approach to Computer Networking"'
Das Buch gibt einen Überblick über Engineering in Computer-Netzwerken. Es werden Grundlagen von Telefon, ATM und Internet aufgezeigt. Es kommen Tools und Techniken zur Sprache: Protocol Layering, Switching, Scheduling, Routing, Error Control, Flow Control ...; Das Buch schließt mit einer Section Practice, in der allgemein angewandte Protokolle und Implementierungen näher beleuchtet werden.


RW97 [RW97]: "`ATM - Infrastruktur für die Hochleistungskommunikation"'
Buch, gut verständlicher Überblick über ATM. Geht u.a. detailliert auf Verkehrssteuerung und statistisches Multiplexen ein.


BGMT98 [BGMT98]: "`Queueing Networks and Markov Chains"'
Eine Einführung in Markovketten. Erklärt Grundlagen von MK mit kontinuerlicher und diskreter Zeit, die Berechnung von steady state und transienter Lösung. Es werden Lösungsalgorithmen für Produkt- und Nicht-Produktformnetze angegeben sowie Werkzeuge zur Lösung, z.B. Mosel vorgestellt. Anwendungsbeispiele komplettieren das Buch.


Stal98 [Stal98]: "`High-Speed Networks: TCP/IP and ATM Design Principles"'
Good textbook about ATM and TCP networks. Figures can be found at http://WilliamStallings.com/HsNet.html


Leh00 [Leh00]: "`Telekommunikation"'
lecture script


Leh01 [Leh01]: "`Modeling and Simulation of Telecommunications Systems and Networks"'
lecture script


Mar99 [Mar99]: "`Traffic Phase Effects in TCP/IP Networks - a Simulation Study"'
Diploma thesis, investigates phase effects.




Atk95 [Atk95]: "`IP Encapsulating Security Payload (ESP)"'
RFC 1827, IPSec, Abstract: This document describes the IP Encapsulating Security Payload (ESP). ESP is a mechanism for providing integrity and confidentiality to IP datagrams. In some circumstances it can also provide authentication to IP datagrams. The mechanism works with both IPv4 and IPv6.


Per96 [Per96]: "`IP Encapsulation within IP"'
RFC 2003, This document specifies a method by which an IP datagram may be encapsulated (carried as payload) within an IP datagram. Encapsulation is suggested as a means to alter the normal IP routing for datagrams, by delivering them to an intermediate destination that would otherwise not be selected by the (network part of the) IP Destination Address field in the original IP header. Encapsulation may serve a variety of purposes, such as delivery of a datagram to a mobile node using Mobile IP.


PF97 [PF97]: "`Why We Don't Know How To Simulate The Internet"'
LBNL-41196, Das Internet ist zu heterogen und ändert sich zu schnell, um es auf traditionelle Weise untersuchen zu können (Links, Protokolle, Application Mix, congestion seen). 2 grundlegende Strategien werden vorgeschlagen: Suche nach Invarianten und Untersuchung des Parameterraums.


DH98 [DH98]: "`Internet Protocol, Version 6 (IPv6) Specification"'
RFC 2460, specifies IPv6


BC99 [BC99]: "`A Performance Evaluation of Hyper Text Transfer Protocols"'
Die Leistung von HTTP/1.0 und HTTP/1.1 werden verglichen. HTTP/1.0 benutzt für jedes Objekt eine TCP-Verbindung, HTTP/1.1 benutzt eine persistente TCP-Verbindung und führt Pipelining durch. HTTP/1.1 ist besser bzgl. Netzwerkperformance, bekommt aber Probleme, wenn der Server den Engpunkt darstellt, speziell bei geringem Speicherausbau. Abhilfe schafft ein früher Verbindungsabbau am Server. Die Ergebnisse werden meßtechnisch erreicht.


MEH00 [MEH00]: "`The Internet: A Gloabal Telecommunications Solution?"'
reviews the different concepts for QoS in IP networks. IntServ, DiffServ, IPv6, RSVP, MPLS. States, that with DiffServ qualitative as well as quantitative services are possible, however additional mechanisms are needed to deploy quantitative services (e.g. route pinning, admission control) which do not belong to DiffServ today. A combination of IntServ (edge) and DiffServ (core) can be used to privide ent-to-end QoS.

Bibliography

GNUTELLA
Gnutella.
software, available at http://gnutella.wego.com.

MOSEL
MOSEL - MOdeling Specification & Evaluation Language.
Software application, (available at http://www4.informatik.uni-erlangen.de/Projects/MOSEL/).

I.321
B-ISDN Protocol reference model and its application.
ITU-T Recommendation I.321, 1991.

UNI3.0
ATM User-Network Interface Specification, Version 3.0.
ATM Forum, technical specification af-uni-0010.002, published by PTR Prentice Hall, 1993.
ISBN 0-13-225863-3.

Fore96
ForeThought Bandwidth Management, version 1.0.
White paper, Fore Systems, 9 pages, 1996.

ATMFTM40
Traffic Management Specification Version 4.0.
ATM Forum, technical specification af-tm-0056.000, 1996.

BegriffeNVT
Nachrichtenverkehrstheorie - Begriffe.
Informationstechnische Gesellschaft im VDE (ITG), Empfehlung 5.2-03 (in german and english), 1997.

ISABEL
ISABEL multiconferencing application.
Software application, (available at http://isabel.dit.upm.es), 1999.

ATMFTM41
Traffic Management Specification, Version 4.1.
ATM Forum, technical specification af-tm-0121.000, 119 pages, 1999.

TM41DIFF
Addendum to TM 4.1: Enhancements to Support IP Differentiated Services and IEEE 802D over ATM Draft.
ATM-Forum, document BTD-TM-DIFF-011.03, (work in progress), 2000.

NAPSTER
Napster.
Software application, (available at http://www.napster.com), 2000.

ns2
ns-2 Simulator.
Software application, (available at http://www.isi.edu/nsnam/ns/), 2000.

REALPLAYER
RealPlayer.
Software application, RealNetworks Inc., (avaiable at http://www.realplayer.com), 2000.

dictionary.com
Dictionary.com.
Online dictionary, (available at http://www.dictionary.com), 2001.

YATS
YATS - Yet Another Tiny Simulator.
Software application, (available at http://www.ifn.et.tu-dresden.de/TK/yats/yats.html), 2001.

AZN98
R.G. Addie, M. Zukerman, and T.D. Neame.
Broadband Traffic Modeling: Simple Solutions to Hard Problems.
IEEE Communications Magazine, 36, 8:88-95, 1998.

Allman98
M. Allman.
On the Generation and Use of TCP Acknowledgements.
to appear in ACM Computer Communications Review, page 18, 1998.

All99
M. Allman.
TCP Byte Counting Refinements.
ACM Computer Communication Review, 29, 3:14-22, 1999.

ABF00
M. Allman, H. Balakrishnan, and S. Floyd.
Enhancing TCP's Loss Recovery Using Limited Transmit.
IETF, RFC 3042, 2001.

AF99
M. Allman and A. Falk.
On the Effective Evaluation of TCP.
ACM Computer Communication Review, 29, 5:59-70, 1999.

AFP98
M. Allman, S. Floyd, and C. Partridge.
Increasing TCP's Initial Window.
IETF, RFC 2414, 1998.

AP99
M. Allman and V. Paxson.
On Estimating End-to-End Network Path Properties.
ACM SIGCOMM'99, pages 263-274, 1999.

APS99
M. Allman, V. Paxson, and W. Stevens.
TCP Congestion Control.
IETF, RFC 2581, 1999.

ABF97
W. Almesberger, J.J.Y. Le Boudec, and T. Ferrari.
Scalable Resource Reservation fot the Internet.
Technical report, EPFL, 1997.

ALPS99
I. Andrikopoulos, A. Liakopoulos, G. Pavlou, and Z. Sun.
Providing Rate Guarantees for Internet Application Traffic Across ATM Networks.
IEEE Communications Surveys, 2, 3:2-13, 1999.

AWP00
I. Andrikopoulos, L. Wood, and G. Pavlou.
A Fair Traffic Conditioner for the Assured Service in a Differentiated Services Internet.
Proceedings of ICC 2000, New Orleans, 2:806-810, 2000.

AMI99
J. Aracil, D. Morato, and M. Izal.
Analysis of Internet Services in IP over ATM Networks.
IEEE Communications Magazine, 37, 12:92-97, 1999.

Armi00
G. Armitage.
MPLS: The Magic Behind the Myths.
IEEE Communications Magazine, 38, 1:124-131, 2000.

AK99
A. Arvidson and P. Karlsson.
On Traffic Models for TCP/IP.
Proceedings of the ITC-16, Elsevier, 1:457-466, 1999.

Atk95
R. Atkinson.
IP Encapsulating Security Payload (ESP).
IETF, RFC 1827, 1995.

Awdu99
D. Awduche.
MPLS and Traffic Engineering in IP Networks.
IEEE Communications Magazine, 37, 12:42-47, 1999.

BBS98
S. Bajaj, L. Breslau, and S. Shenker.
Is Service Priority Useful in Networks.
SIGMETRICS '98, pages 66-77, 1998.

Bakre94
A. Bakre and B.R. Badrinath.
I-TCP: Indirect TCP for Mobile Hosts.
Technical report, Rutgers University, Piscataway, NJ 08855, Dept. of Computer Science, DCS-TR-314, 1994.

BPK99
H. Balakrishnan, V.N. Padmanabhan, and R.H. Katz.
The Effects of Asymmetry on TCP Performance.
Technical report, University of California at Berkeley, Computer Science Division, 1999.

BAD00
C. Barakat, E. Altman, and W. Dabbous.
On TCP performance in a Heterogeneous Network: A Survey.
IEEE Communications Magazine, 38, 1:40-46, 2000.

BC99
P. Barford and M. Crovella.
A Performance Evaluation of Hyper Text Transfer Protocols.
Proceedings of ACM SIGMETRICS '99, pages 188-197, 1999.

BP97
D. Basak and S. Pappu.
GFR Implementation Alternatives with Fair Buffer Allocation Schemes.
ATM Forum, document 97-0528, 1997.

Bau97
M. Baumann.
Einige Gedanken und Ergebnisse zu TCP über UBR/EPD/ABR.
Technical report, Technische Universität Dresden, Communications Laboratory, Chair for Telecommunications, 1997.

YATSUSER
M. Baumann.
YATS - Yet Another Tiny Simulator, User's Manual for Version 0.3.
TU-Dresden, Communications Laboratory, (available at http://www.ifn.et.tu-dresden.de/TK/yats/yats.html), 2000.

BMO+98
M. Baumann, T. Müller, W. Ooghe, A. Santos, et al.
Multi-Layer Modelling of a Multi-Media Application.
Proc. 4th International Conference on Broadband Communications '98, Stuttgart, pages 565-576, 1998.

BPS99
J. Bennet, C. Partridge, and N. Shectman.
Packet Reordering is Not Pathological Network Behavior.
IEEE/ACM Transactions on Networking, 7, 6:789-798, 1999.

IDBern99
Y. Bernet et al.
Interoperation of RSVP/Intserv and Diffserv Networks, 1999.

BSJ99
P. Bhaniramka, W. Sun, and R. Jain.
Quality of Service using Traffic Engineering over MPLS: An Analysis.
Technical report, Ohio State University, Columbus, USA, CIS/OSU, (avail. at http://www.cis.ohio-state.edu/jain/papers.html), 1999.

BLN99
V. Bhargavan, S. Lu, and T. Nandagopal.
Fair Queuing in Wireless Networks: Issues and Approaches.
IEEE Network, pages 44-53, 1999.

LB00
E.W. Biersach and A. Legout.
Pathological Behaviors for RLM and RLC.
Proceedings of NOSSDAV '2000, page 9, 2000.

Bla00
D. Black.
Differentiated Services and Tunnels.
IETF, RFC 2983, 2000.

BBCD+98
S. Blake, D. Black, M. Carlson, E. Davies, Z. Wang, and W. Weiss.
An Architecture for Differentiated Services.
IETF, RFC 2475, 1998.

BGMT98
G. Bolch, S. Greiner, H. de Meer, and K. Trivedi.
Queueing Networks and Markov Chains.
John Wiley & Sons, Inc., 1998.
ISBN 0-471-19366-6.

Bona98
O. Bonaventure.
Integration of ATM under TCP/IP to provide services with guaranteed minimum bamdwidth.
Ph.D. thesis, University de Liege, Faculte des Sciences Appliquees, 1998.

BC00
O. Bonaventure and S. de Cnodder.
A rate adaptive shaper for differentiated services.
IETF, RFC 2963, 2000.

BCCD+98
B. Braden, D. Clark, J. Crowcroft, B. Davie, et al.
Recommendations on Queue Management and Congestion Avoidance in the Internet.
IETF, RFC 2309, 1998.

Bra89
R. Braden.
Requirements for Internet Hosts - Communication Layers.
IETF, RFC 1122, 1989.

BCS94
R. Braden, D. Clark, and S. Shenker.
Integrated Services in the Internet Architecture: an Overview.
IETF, RFC 1633, 1994.

BZB+97
R. Braden, L. Zhang, S. Berson, S. Herzog, and S. Jamin.
Resource ReSerVation Protocol (RSVP) - Version 1, Functional Specification.
IETF, RFC 2205, 1997.

BP95
L. Brakmo and L. Peterson.
TCP Vegas: End to End Congestion Avoidance on a Global Internet.
IEEE Journal on Selected Areas in Communications, 13, 8:1465-1480, 1995.

BM98
M. Breugs and T. Magedanz.
Mobile Agents - Enabling Technology for Active Intelligent Network Implementation.
IEEE Network, 12, 3:53-60, 1998.

BWBM98
U. Briem, E. Wallmeier, C. Beck, and F. Matthiesen.
Traffic Management for an ATM Switch with Per-VC Queueing: Concept and Implementation.
IEEE Communications Magazine, 36, 1:88-93, 1998.

BS89
J.N. Bronstein and K.A. Semendjajew.
Taschenbuch der Mathematik.
BSB B. G. Teubner Verlagsgesellschaft, 24. Auflage (in german), 1989.
ISBN 3-322-00259-4.

CWZ00
Z. Cao, Z. Wang, and E. Uegura.
Rainbow Fair Queueing: Fair Bandwidth Sharing Without Per-Flow State.
IEEE Infocom '2000, Tel-Aviv, Israel, 2000.

CFE99
P.C. Charalambos, V.S. Frost, and J.B. Evans.
Performance Evaluation of TCP Extensions on ATM over High Bandwidth Delay Product Networks.
IEEE Communications Magazine, 37, 7:57-63, 1999.

Char2000
J. Charzinski.
Internet Client Traffic Measurement and Characterisation Results.
Proc. of ISSLS2000, Stockholm, Sweden, (avail. at www.issls-council.org/issls00.htm), 10 pages, 2000.

Char00c
J. Charzinski.
Internet Client Traffic Measurements and Characterization Results, 2000.

Char00b
J. Charzinski.
Web Performance in Practice - Why We are Waiting.
International Journal of Electronics and Communications (AEÜ), Urban & Fischer, ISSN 1434-8411, 55, 1:37-45, 2001.

Char00a
J. Charzinsky.
HTTP/TCP connection and flow characteristics.
Performance Evaluation, 42:149-162, 2000.

CP99
S. Chen and K. Park.
An Architecture for Noncooperative QoS Provision in Many-Switch Systems.
IEEE Infocom '99, 1999.

CLWS96
T. Chen, S. Liu, D. Wang, V. Samalam, et al.
Monitoring and Control of ATM Networks Using Special Cells.
IEEE Network, 10, 5:23-38, 1999.

CO99
T.M. Chen and T.H. Oh.
Reliable Services in MPLS.
IEEE Communications Magazine, 37, 12:58-62, 1999.

CG00
N. Ciulli and S. Giordano.
Implementation and Experimental Analysis of IP Schedulers in an IntServ-over-ATM Test-bed.
13th ITC specialists seminar, Monterey, USA, pages 25/1-25/10, 2000.

CF98
D.D. Clark and W. Fang.
Explicit Allocation of Best Effort Packet Delivery Service.
IEEE/ACM Transactions on Networking, 6, 4:362-374, 1998.

CGE98
J.A. Cobb, M.G. Gouda, and A. El-Nahas.
Time-Shift Scheduling - Fair Scheduling of Flows in High-Speed Networks.
IEEE/ACM Transactions on Networking, 6, 3:274-285, 1998.

CH00
R. Cohen and Y. Hamo.
Balanced Packet Discard for Improving TCP Performance in ATM Networks.
IEEE Infocom '2000, 2000.

CG99
G. Cope and N. Garnham.
QoS Control and Interworking of Overbooked Elastic and Brittle ATM Traffic.
Proceedings of the ITC-16, Elsevier, 1:205-216, 1999.

Costeux99
Jean-Laurent Costeux.
Fluid Analysis of a TCP Connection over ABR.
Proceedings of the ITC-16, Elsevier, 1:251-262, 1999.

Cra+98
E. Crawley et al.
A Framework for Integrated Services and RSVP over ATM.
IETF, RFC 2382, 1998.

Crovella96
M.E. Crovella and A. Bestravos.
Self-Similarity in World Wide Web Traffic - Evidence and Possible Causes.
ACM Sigmetrics Conf. on Measurement and Modelling of Computer Systems, Philadeliphia, USA, pages 160-169, 1996.

CO98
J. Crowcroft and P. Oechslin.
Differentiated End-to-End Internet Services using a Weighted Proportional Fair Sharing TCP.
ACM Computer Communication Review, 28, 3:53-69, 1998.

Dahl99
D. Dahlmeyer.
Betrieb und Wartung von ATM-Netzen mit OAM-Zellen.
NTZ, 6:62-65, 1999.

Cnod00
S. de Cnodder.
Rade Adaptive Shapers for Data Traffic in DiffServ Networks.
Networld + Interop, 2000.

CEP00
S. de Cnodder, O. Elloumi, and K. Pauwels.
RED Behavior with Different Packet Sizes.
IEEE Symposium on Computers and Communications (ISCC), 2000.

DH98
S. Deering and R. Hinden.
Internet Protocol, Version 6 (IPv6) Specification.
IETF, RFC 2460, 1998.

DKS90
A. Demers, S. Keshav, and S. Shenker.
Analysis and Simulation of a Fair Queueing Algorithm.
Journal of Internetworking: Research and Experience, 1:3-26, 1990.

DGP99
E. Desmet, D. Gastaud, and G.H. Petit.
Quality of Service in the Internet.
Alcatel Telecommunications Review, 2nd Quarter, pages 146-150, 1999.

DR99
C. Dovrolis and P. Ramanathan.
A Case for Relative Differentiated Services and the Proportional Differentiation Model.
IEEE Network, 13, 5:26-34, 1999.

DS99
C. Dovrolis and D. Stiliadis.
Relative Differentiated Services in the Internet: Issues and Mechanisms.
Proceedings of ACM SIGMETRICS '99, pages 204-205, 1999.

Down99
A. Downey.
Using pathchar to estimate Internet link characteristics.
Proceedings of ACM SIGMETRICS '99, pages 222-223, 1999.

EHMV+00
G. Eichler, H. Hussmann, G. Mamais, I. Venieris, et al.
Implementing Integrated and Differentiated Services for the Internet with ATM Networks: A Practical Approach.
IEEE Communications Magazine, 38, 1:132-141, 2000.

EA97
O. Elloumi and H. Afifi.
RED Algorithm in ATM Networks.
Technical report, Ecole Nationale Superieure des Telecommunications de Bretagne, Networks and Multimedia Department, 1997.

ECP99
O. Elloumi, S. de Cnodder, and K. Pauwels.
Usefulness of three drop precedences in Assured Forwarding service.
IETF, draft, draft-elloumi.diffserv-threevstwo-00.txt, (work in progress), 1999.

Theo98
T. Faber.
ACC: Using Active Networking to Enhance Feedback Congestion Control Mechanisms.
IEEE Network, 12, 3:61-65, 1998.

FJRGV99
S. Fahmry, R. Jain, S. Rabie, R. Goyal, and B. Vandalore.
Quality of Service for Internet Traffic over ATM Service Categories.
Journal of Computer Communications, Special issue on enterprise networks, 1999.

Fahm98
S. Fahmy and R. Jain.
ABR Flow Control for Multipoint Connections.
IEEE Network, pages 6-7, 1998.

ATMF990403
S. Fahmy, R. Jain, B. Vandalore, and R. Goyal.
A Framework for Virtual Channel onto Virtual Path Multiplexing in ATM-ABR.
ATM Forum, document 990403, 15 pages, 1999.

FF96
K. Fall and S. Floyd.
Simulation-based Comparisons of Tahoe, Reno, and SACK TCP.
ACM Computer Communication Review, 26, 3:5-21, 1996.

FSN00
W. Fang, N. Seddigh, and B. Nandy.
A Time Sliding Window Three Colour Marker (TSWTCM).
IETF, RFC 2859, 2000.

FPLZ98
A. Fei, G. Pei, R. Liu, and L. Zhang.
Measurements on Delay and Hop-Count of the Internet.
GLOBECOMM '98, available at http://irl.cs.ucla.edu/papers/internet-measurement.ps.gz, 1998.

FRC98
A. Feldmann, J. Rexford, and R. Caceres.
Efficient Policies for Carrying Web Traffic over Flow-Switched Networks.
IEEE/ACM Transaction on Networking, 6, 6:673-685, 1998.

Feng99
W.C. Feng.
Improving Internet Congestion Control and Queue Management Algorithms.
Ph.D. thesis, University of Michigan, USA, Dept. of Computer Science and Engineering, 1999.

FKSS97
W.C. Feng, D. Kandlur, D. Saha, and K. Shin.
Techniques for Eliminating Packet Loss in Congested TCP/IP Networks.
Technical report, CSE-TR-349-97, Univ. of Michigan, USA, ftp.eecs.umich.edu/techreports/cse/1997/CSE-TR-349-97.ps.gz, 1997.

FKSS99b
W.C. Feng, D. Kandlur, D. Saha, and K. Shin.
Adaptive Packet Marking for Maintaining Ent-to-End Throughput in a Differentiated-Services Internet.
IEEE/ACM Transactions on Networking, 7, 5:685-697, 1999.

FKSS99
W.C. Feng, D. Kandlur, D. Saha, and K.G. Shin.
Understanding and Improving TCP Performance Over Networks with Minimum Rate Guarantees.
IEEE Transactions on Networking, 7, 2:173-186, 1999.

FKR00
A. Feroz, S. Kalyanaraman, and A. Rao.
A TCP-friendly traffic marker for IP differentiated services.
Eighth International Workshop on Quality of Service (IWQoS), Pittsburgh, pages 138-147, 2000.

Floyd95
S. Floyd.
TCP and Successive Fast Retransmits.
Technical report, Lawrence Berkeley Laboratory, 1995.

Flo00
S. Floyd.
Congestion Congrol Principles.
IETF, RFC 2914, 2000.

FF99
S. Floyd and K. Fall.
Promoting the Use of End-to-End Congestion Control in the Internet.
IEEE/ACM Transactions on Networking, 7, 4:458-472, 1999.

FH99
S. Floyd and T. Henderson.
The NewReno Modification to TCP's Fast Recovery Algorithm.
IETF, RFC 2582, 1999.

FJ92
S. Floyd and V. Jacobson.
On Traffic Phase Effects in Packet-Switched Gateways.
Journal of Internetworking: Research and Experience, 3, No. 3:115-156, 1992.

FJ93
S. Floyd and V. Jacobson.
Random Early Detection Gateways for Congestion Avoidance.
IEEE/ACM Transactions on Networking, 1, 4:397-413, 1993.

FBC99
J. Färber, S. Bodamer, and J. Charzinski.
Statistical Evaluation and Modelling of Internet Dial-Up Traffic.
Proceedings of the SPIE, Boston, 19-22, 1999.

Amit98
A. Garg and A. Nanavati.
Routers Mechanisms for Congestion Avoidance.
Technical report, University of California, Los Angeles, (available at http://irl.cs.ucla.edu/papers/amit-report.ps.gz), 1998.

GB98
M. Garrett and M. Borden.
Interoperation of Controlled-Load Service and Guaranteed Service with ATM.
IETF, RFC 2381, 1998.

Ghani99
N. Ghani.
Enhanced TCP/IP ACK Pacing for High Speed Networks.
Proccedings of the ITC-16, Elsevier, 1:489-498, 1999.

GD99
N. Ghani and S. Dixit.
TCP/IP Enhancements for Satellite Networks.
IEEE Communications Magazine, pages 64-72, 1999.

GJFA+99
A. Ghanwani, G. Jamoussi, D. Fedyk, P. Ashwood-Smith, et al.
Traffic Engineering Standards in IP Networks Using MPLS.
IEEE Communications Magazine, 37, 12:49-53, 1999.

GM99
P. Giacomazzi and L. Musumeci.
Transport of IP Controlled-Load Service over ATM Networks.
IEEE Network, 13, 1:36-47, 1999.

Gol94
J. Golestani.
A Self-Clocked Fair Queueing Scheme for Broadband Applications.
Proceedings of Infocom '94, pages 636-646, 1994.

GMJ99
M. Goyal, P. Misra, and R. Jain.
Effect of Number of Drop Precedences in Assured Forwarding.
Global Telecommunications Conference, GLOBECOM '99, Brazil, Dezember, 5-9,, 1a:188-193, 1999.

Goyal99
R. Goyal.
Traffic Management for TCP/IP over Asynchronous Transfer Mode (ATM) Networks.
Ph.D. thesis, Ohio State University, CIS, USA, (available at http://www.cis.ohio-state.edu/jain/papers/rohitphd.htm), 1999.

GJFV98
R. Goyal, R. Jain, S. Fahmy, and B. Vandalore.
Buffer Management for the GFR Service.
ATM Forum, document 98-0405, 15 pages, 1998.

GJF+97
R. Goyal, R. Jain, S. Fahmy, B. Vandalore, and S. Kalyanaraman.
Simulation Experiments with Guaranteed Frame Rate for TCP/IP Traffic.
ATM Forum, document 97-0607, 12 pages, 1997.

GH96
R. Guerin and J. Heinanen.
UBR+ Service Category Definition.
ATM Forum, document 96-1598, 9 pages, 1996.

HPF00
M. Handley, J. Padhye, and S. Floyd.
TCP Congestion Window Validation.
Internet Draft draft-handley-tcp-cwv-02.txt (work in progress), 2000.

HBWW99
J. Heinanen, F. Baker, W. Weiss, and J. Wroclawski.
Assured Forwarding PHB Group.
IETF, RFC 2597, 1999.

HG99a
J. Heinanen and R. Guerin.
A Single Rate Three Color Marker.
IETF, RFC 2697, 1999.

HG99b
J. Heinanen and R. Guerin.
A Two Rate Three Color Marker.
IETF, RFC 2698, 1999.

Heiss99
H. Heiss.
QoS results for the GFR ATM service.
Proccedings of the ITC-16, Elsevier, 1:227-236, 1999.

HV98
F. Hellstrand and A. Veres.
Simulation of TCP/IP Router Traffic over ATM using GFR and VBR.3.
ATM Forum, document 98-0087, 12 pages, 1998.

Hilg99
U. Hilgers.
Differentiated Services - Konzept und erste Erfahrungen.
MMB'99, Trier, Report No. 99-17, pages 55-59, 1999.

Jaco95
V. Jacobson.
Congestion Avoidance and Control.
Computer Communication Review, a publication of ACM SIGCOMM, 25, 1:158-173, 1995.

JBB92
V. Jacobson, R. Braden, and D. Bormann.
TCP Extensions for High Performance.
IETF, RFC 1323, 1992.

JNP99
V. Jacobson, K. Nichols, and K. Poduri.
An Expedited Forwarding PHB.
IETF, RFC 2598, 1999.

Jain90
R. Jain.
Congestion Control in Computer Networks: Issues and Trends.
IEEE Network, pages 24-30, 1990.

Jain91
R. Jain.
The Art of Computer Systems Performance Analysis: Techniques for Experimental Design, Measurement, Simulation, and Modeling.
John Wiley & Sons, Inc., 1991.
ISBN 0-471-50336-3.

Jain92
R. Jain.
Myths About Congestion Management in High-Speed Networks.
Internetworking: Research and Experience, 3:101-113, 1992.

ATMF940881
R. Jain.
Fairness: How to Measure Quantitatively?
ATM Forum, document 94-0881, 1994.

Jain97
R. Jain.
A Survey of Scheduling Methods.
Slides, Ohio State University, CIS, USA, (avail. at http://www.cis.ohio-state.edu/jain/talks/sched.htm), 1997.

ATMF990045
R. Jain, A. Duressi, and G. Babic.
Throughput Fairness Index: An Explanation.
ATM Forum, document 99-0045, 1999.

ERICA96
R. Jain, S. Kalyanaraman, R. Goyal, et al.
ERICA Switch Algorithm: A Complete Description.
ATM Forum, document 96-1172, 44 pages, 1996.

Jain84
R.K. Jain, D.M.C. Chiu, and W.R. Hawe.
A Quantitative Measure of Fairness and Discrimination for Resource Allocation in Shared Computer System.
ACM Transactions on Computer Systems, page 37, 1984.

JLN98
L. Jaussi, M. Lorang, and J. Nelissen.
A Detailed Experimental Performance Evaluation on TCP over UBR.
Proceedings 1st IEEE International Conference on ATM, ICATM'98, Colmar, France, 1998.

KVR98
L. Kalampoukas, A. Varma, and K.K. Ramakrishnan.
Two-Way TCP Traffic over Rate Controlled Channels: Effects and Analysis.
IEEE/ACM Transactions on Networking, 6, 6:729-743, 1998.

KJFG98
S. Kalyanaraman, R. Jain, S. Fahmy, and R. Goyal.
Performance and Buffering Requirements of Internet Protocols over ATM ABR and UBR Services.
IEEE Communications Magazine, 36, 6:152-157, 1998.

KR99
M. Kara and M.A. Rahin.
Comparison of Service Disciplines in ATM Networks Using a Unified QoS Metric.
Proceedings of the ITC-16, Elsevier, 2:1137-1146, 1999.

KSC91
M. Katevenis, S. Sidiropoulos, and C. Courcoubetis.
Weighted round-robin cell multiplexing in a general-purpose ATM switch chip.
IEEE Journal on Selected Areas in Communications, 9, 8:1265 - 1279, 1991.

Kesh97
S. Keshav.
An Engineering Approach to Computer Networking.
Addison-Wesley, 1997.
ISBN 0-201-63442-2.

KS99
S. Köhler and U. Schäfer.
Performance Comarison of Different Class-and-Drop Treatment of Data and Acknowledgements in DiffServ IP Networks.
Technical report, University of Würzburg, Dep. of Computer Science, Lehrstuhl für Informatik III, Report No. 237, 1999.

KLT98
H. Kim, W.E. Leland, and S.E. Thomson.
Evaluation of Bandwidth Assurance Service using RED for Internet Service Differentiation.
Technical report, Bell Communications Research, (available at ftp://ftp.bellcore.com/pub/world/hkim/assured.ps.Z), 1999.

ATMF980290
L. Klampoukas, A. Varma, and K. Ramakrishnan.
Improving TCP/IP Performance over ATM's ABR Service with Explicit Window Adaption.
ATM Forum, document 98/0290, 14 pages, 1998.

Kori92
Y. Korilis and A. Larzar.
Why is Flow Control Hard: Optimality, Fairness, Partial and Delayed Information.
Technical report, Center for Telecommunications Research, Columbia University, 1992.

Krun99
M. Krunz.
Bandwidth Allocation Strategies for Transporting Variable-Bit-Rate Video Traffic.
IEEE Communications Magazine, 37, 1:40-46, 1999.

KL98
L.A. Kulkarni and S.Q. Li.
Performance Analysis of a Rate-Based Feedback Control Scheme.
IEEE/ACM Transactions on Networking, 6, 6:797-810, 1998.

KLS98
V.P. Kumar, T.V. Lakshman, and D. Stiliadis.
Beyond Best Effort: Router Architectures for the Differentiated Services of Tomorrow's Internet.
IEEE Communications Magazine, 36, 5:152-164, 1998.

KBC94
H.T. Kung, T. Blackwell, and A. Chapman.
Credit-Based Flow Control for ATM Networks: Credit Update Protocol, Adaptive Credit Allocation, and Statistical Multiplexing.
ACM SIGCOMM '94, 8:101-114, 1994.

KM95
H.T. Kung and R. Morris.
Detailed Behavior of TCP over ATM.
GLOBECOM ´95, page 6, 1995.

LB99
M.A. Labrador and S. Banerjee.
Packet Dropping Policies for ATM and IP Networks.
IEEE Communications Surveys, 3rd quarter, 2, 3:2-14, 1999.

Laub94
M. Laubach.
Classical IP and ARP over ATM.
IETF, RFC 1577, 1994.

LH98
M. Laubach and J. Halpern.
Classical IP and ARP over ATM.
RFC 2225, 1998.

Lee98
B.J. Lee.
GFR Dimensioning for TCP Traffic.
ATM Forum, document 98-0271, 11 pages, 1998.

Lefel96
C. Lefelhocz, B. Lyles, S. Shenker, and L. Zhang.
Congestion Control for Best-Effort Service: Why We Need a New Paradigm.
IEEE Network, 1/2:10-19, 1996.

Leh00
R. Lehnert.
Telekommunikation.
Lecture script (in german), TU-Dresden, Communications Laboratory, Chair for Telecommunications, 121 pages, 2000.

Leh01
R. Lehnert.
Modeling and Simulation of Telecommunications Systems and Networks.
Lecture script, TU-Dresden, Communications Laboratory, Chair for Telecommunications, 130 pages, 2001.

LSTIS95
H. Li, K.Y. Siu, and H.Y. Tzeng.
TCP Performance over ABR and UBR Services in ATM.
submitted to IPCCC'96, 1995.

Li99
T. Li.
MPLS and the Evolving Internet Architexture.
IEEE Communications Magazine, 37, 12:38-41, 1999.

LK98
D. Lin and H.T. Kung.
TCP Fast Recovery Strategiew: Analysis and Improvements.
IEEE INFOCOM´98, page 9, 1998.

LM97
D. Lin and R. Morris.
Dynamics of Random Early Detection.
Proceedings of ACM SIGCOMM '97, 27, 4:127-137, 1997.

LJ99
C. Liu and R. Jain.
Improving Explicit Congestion Notification with the Mark-Front Strategy.
Computer Networks, 35, 2/3:185-201, 2001.

LHL99
D. Liu, F. Huebner, and Y. Levy.
A Hierarchical Multi-Class Traffic Model for Data Networks.
Proceedings of the ITC-16, Elsevier, 2:1221-1229, 1999.

LK00
R. Ludwig and R.H. Katz.
The Eifel Algorithm: Making TCP Robust Against Spurious Retransmissions.
ACM SIGCOMM, Computer Communications Review, 30, 1:30-36, 2000.

LRKOJ99
R. Ludwig, B. Rathonyi, A. Konrad, K. Oden, and A. Joseph.
Multi-Layer Tracing of TCP over a Reliable Wireless Link.
Proceedings of ACM SIGMETRICS '99, pages 144-154, 1999.

Loewe96
G. Löwe.
Abhängigkeit der Dienstgüteparameter bei Datenübertragung mittels TCP/IP in ATM-Netzen.
Master's thesis, (in german), Technische Univ. Dresden, Communications Laboratory, Chair for Telecommunications, 1996.

MR91
A. Mankin and K. Ramakrishnan.
Gateway Congestion Control Survey.
IETF, RFC 1254, 1991.

Manth97
S. Manthorpe and J.Y. Boudec.
A comparison of ABR and UBR to support TCP traffic.
Technical report, No. 97/224 EPFL, Lausanne, Switzerland, (available at http://icawww1.epfl.ch/PS_files/publications.html), 1997.

Manth95
S. Manthorpe and X. Garcia.
TCP Performance Over ATM Based LAN Interconnection Services.
Procedings of Interop '95 Engineers Conference, 1995.

MBLM99
M.A. Marsan, A. Bianco, E. Leonardi, and L. Milia.
A QoS-Aware Scheduling Algorithm for Input Queued Switches.
SPECTS 99, pages 55-61, 1999.

Mar99
M. Martinez.
Traffic Phase Effects in TCP/IP Networks - a Simulation Study.
Master's thesis, Technische Universität Dresden (TUD), Communications Laboratory, Chair for Telecommunications, 1999.

MR99
L. Massoulie and J.W. Roberts.
Arguments in favour of admission control for TCP flows.
Proceedings of the ITC-16, Elsevier, 1:33-44, 1999.

MMFR96
M. Mathis, J. Mahdavi, S. Floyd, and A. Romanow.
TCP Selective Acknowledgement Options.
IETF, RFC 2018, 1996.

MSMO97
M. Mathis, J. Semke, J. Mahdavi, and T. Ott.
The Macroscopic Behavior of the TCP Congestion Aviodance Algorithm.
ACM SIGCOMM Computer Communication Review, 27, 3:67-82, 1997.

MEH00
L. Mathy, C. Edwards, and D. Hutchison.
The Internet: A Global Telecommunications Solution?
IEEE Network, 4:46-57, 2000.

MC00
S. McCreary and K. Claffy.
Trends in Wide Area IP Traffic Patterns: A View from Ames Internet eXchange.
13th ITC Specialist Seminar "IP traffic meas., modeling and management", Monterey, USA, pages 1/1-1/11, 2000.

MH00
A. Mena and J. Heidemann.
An Empirical Study of Real Audio Traffic.
Proceedings of the IEEE Infocom, Tel-Aviv, Israel, pages 101-110, 2000.

Mue00a
T. Müller.
Guaranteed QoS for TCP flows in ATM-based DiffServ Networks.
IEEE, 1st European Conference on Universal Multiservice Networks, ECUMN'2000, Colmar, France, pages 158-163, 2000.

Mue00b
T. Müller.
QoS Guarantees for TCP Flows in Differentiated-Services Networks.
First Polish-German Teletraffic Symposium (PGTS2000), Dresden, Germany, pages 65-72, 2000.

MHL99
T. Müller, H. Hrasnica, and R. Lehnert.
Influence of the ATM Service Category on TCP Performance in Large Network.
Proceedings of the SPIE, Boston, USA, 1999.

MS99
T. Müller and J. Schüler.
Traffic Phase Effects in Realistic IP Networks.
MMB'99, published as report 99-17, University of Trier, pages 49-53, 1999.

Mold95
K. Moldeklev and P. Gunningberg.
How a Large ATM MTU Causes Deadlocks in TCP Data Transfers.
IEEE/ACM Transactions On Networking, 3, 4:409-422, 1995.

MG99
W. Monin and A. Gravey.
Intelligent Credit Management for a Credit Based Congestion Control Procedure.
Proceedings of the ITC-16, Elsevier, 1:415-424, 1999.

ML00
R. Morris and D. Lin.
Variance of Aggregated Web Traffic.
IEEE Infocom '2000, 2000.

NC99
J. Nelissen and S. de Cnodder.
A Versatile RED-based Buffer Management Mechanism for the Efficient Support of Internet Traffic.
Proc. of SPIE, Internet II: Quality of Service and Future Directions, Boston, USA, 3842:57-68, 1999.

NSSZ99
E. Ng, D. Stephens, I. Stoica, and H. Zhang.
Supporting Best-Effort Traffic With Fair Service Curve.
Proceedings of ACM SIGMETRICS '99, 1:218-219, 1999.

Nic01
K. Nichols.
Definition of Differentiated Services Per Domain Behaviors and Rules for their Specification.
IETF draft, draft-ietf-diffserv-pdb-def-03 (work in progress), 2001.

NBBB98
K. Nichols, S. Blake, F. Baker, and D. Black.
Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers.
IETF, RFC 2474, 1998.

NJZ97
K. Nichols, V. Jacobson, and L. Zhang.
A Two-bit Differentiated Services Architecture for the Internet, 1997.

Ott97
T.J. Ott and N. Aggarwal.
TCP over ATM: ABR or UBR?
Proceedings of ACM SIGMETRICS '97, pages 52-63, 1997.

PFTK98
J. Padhye, V. Firoiu, D. Towsley, and J. Kurose.
Modeling TCP Throughput: A Simple Model and its Empirical Validation.
IEEE SIGCOMM '98, Vancouver, Canada, 1998.

PKT99
J. Padhye, J. Kurose, and D. Towsley.
A TCP-Friendly Rate Adjustment Protocol for Continuous Media Flows over Best Effort Networks.
Proceedings of ACM SIGMETRICS '99, pages 220-221, 1999.

PPP00
R. Pan, B. Prabhakar, and K. Psounis.
Choke - a stateless active queue management scheme for approximating fair bandwidth allocation.
IEEE Infocom '2000, Tel-Aviv, Israel, page 10, 2000.

PG92
A.K. Parekh and R.G. Gallager.
A generalized processor sharing approach to flow control in integrated services networks - the single node case.
IEEE/ACM Transactions on Networking, 1, 3:344-357, 1993.

Par91
A.K.J. Parekh.
A Generalized Processor Sharing Approach to Flow Control in Integrated Services Networks.
Ph.D. thesis, Massachusetts Institute of Tech (MIT), Cambridge, USA, Dep. of Electr. Eng. and Com. Science, LIDS-TH-2089, 1991.

Par98
C. Patridge et al.
A 50-Gb/s IP Router.
IEEE/ACM Transactions on Networking, 6, 3:237-248, 1998.

Pax97b
V. Paxson.
Automated Packet Trace Analysis of TCP Implementations.
ACM SIGCOMM ´97, 27, 4:167-179, 1997.

Pax97
V. Paxson.
End-to-End Internet Packet Dynamics.
ACM Computer Communication Review, 27, 4:139-152, 1997.

PaxPhd97
V. Paxson.
Measurement and Analysis of End-to-End Internet Dynamics.
Ph.D. thesis, University of California, Berkeley, USA, Computer Science Division, 386 pages, LBNL-40319, 1997.

PA00
V. Paxson and M. Allman.
Computing TCP's Retransmission Timer.
IETF, RFC 2988, 2000.

PF97
V. Paxson and S. Floyd.
Why We Don't Know How To Simulate The Internet.
Technical report, Proceedings of the Simulation Conference, 1997.

Per96
C. Perkins.
IP Encapsulation within IP.
IETF, RFC 2003, 1996.

Plon00
D. Plonka.
Internet Traffic Flow Size Analysis.
Technical report, University of Wisconsin - Madison, USA, (available at http://net.doit.wisc.edu/data/flow/size/), 2000.

Pos80
J. Postel.
User Datagram Protocol.
IETF, RFC 768, 1980.

RFC793
J. Postel.
Transmission Control Protocol.
IETF, RFC 793, 1981.

PKD00
S. Puangpronpitag, M. Kara, and K. Djemame.
A Performance Evaluation of Buffer Management and Scheduling for ATM-GFR Using TCP.
Proc. of the 8th IFIP Workshop on Perf. Modelling and Evaluation of ATM and IP Networks, Ilkley, pages 71/1-71/14, 2000.

RF99
K. Ramakrishnan and S. Floyd.
A Proposal to add Explicit Congestion Notification (ECN) to IP.
IETF, RFC2481, 1999.

RW97
E. Rathgeb and E. Wallmeier.
ATM - Infrastruktur für die Hochleistungskommunikation.
Springer Verlag (in german), 1997.
ISBN 3-540-60370-0.

Rein99
D. Reiniger et al.
Soft QoS Control in the WATMnet Broadband Wireless System.
IEEE Personal Communications, pages 24-43, 1999.

Rie00
U. Riedel.
Simulative Untersuchung eines Verfahrens zur Dienstgütegarantie in DiffServ IP Netzen.
Master's thesis, (in german), Technische Univ. Dresden, Communications Laboratory, Chair for Telecommunications, 2000.

RobertsCIF97
D.L.G. Roberts.
Cells In Frames - ATM with Variable Length Packets.
available at: http://www.ziplink.net/lroberts/cells-In-Frames.html, 1997.

Roberts97
D.L.G. Roberts.
Explicit Rate Flow Control - A 100 Fold Improvement over TCP.
available at: http://www.ziplink.net/lroberts/ERvsTCP.html, 1997.

Rob97
J.W. Roberts.
Virtual Spacing for Flexible Traffic Control.
International Journal of Communication Systems, 7:307-318, 1997.

RLMA99
G. Rogers, R. Liu, M. Minhazuddin, and J. Argyros.
Matching Differentiated Services PHBs to ATM Service Categories - the AF to VBR Case.
IEEE Lanman´99 Workshop, 21-24, page 13, 1999.

RF95
A. Romanow and S. Floyd.
Dynamics of TCP Traffic over ATM Networks.
IEEE Journal on Selected Areas in Communications (JSAC), 13, 4:633-641, 1995.

RVC01
E. Rosen, A. Viswanathan, and R. Callon.
Multiprotocol Label Switching Architecture.
IETF, RFC 3031, 2001.

Ros96
J. Rosenberg.
Resource Allocation on the Cell Level: Link Scheduling and Buffer Management.
Technical report, Lucent Technologies, Bell Laboratories, Holmdel, USA, (available at www.columbia.edu/gd47/writeup.html), 1996.

RBL99
V. Rosolen, O. Bonaventure, and G. Leduc.
A RED discard strategy for ATM networks and its performance evaluation with TCP/IP traffic.
Computer Communication Review, a publication of ACM SIGCOMM, ISSN 0146-4833, 29, 3:23-43, 1999.

ST99
Z. Sahinoglu and S. Tekinay.
On Multimedia Networks: Self-Similar Traffic and Network Performance.
IEEE Communications Magazine, 37, 1:48-52, 1999.

STK99
S. Sahu, D. Towsley, and J. Kurose.
A Quantitative Study of Differentiated Services for the Internet.
Global Telecommunications Conference, GLOBECOM '99, 3:1808-1817, 1999.

Sama99
N.K.G. Samaraweera.
Return Link Optimization for Internet Service Provision Using DVB-S Networks.
ACM SIGCOMM, Computer Communications Review, 29, 3:4-13, 1999.

ATMF980152
R. Satyavolu, K. Duvedi, and S. Kalyanaraman.
Explicit rate control of TCP applications.
ATM Forum, document 98-0152, 16 pages, 1998.

SCWA99
S. Savage, N. Cardwell, D. Wetherall, and T. Anderson.
TCP Congestion Control with a Misbehaving Receiver.
ACM Computer Communication Review, 29, 5:71-78, 1999.

SHS97
H. Sawashima, Y. Hori, and H. Sunahara.
Characteristics of UDP Packet Loss: Effect of TCP Traffic.
INET '97, available at http://www.chg.ru/inet97/F3/F3_1.HTM, 1997.

Schueler98
J. Schüler.
Fairness bei TCP-Datenübertragung in ATM-Netzwerken.
Master's thesis, (in german), Technische Univ. Dresden, Communications Laboratory, Chair for Telecommunications, 1998.

SM99
J. Schüler and T. Müller.
Dynamics of TCP in congested ATM multiplexers.
MMB'99, published as report 99-17, University of Trier, pages 9-11, 1999.

SBGF97
R. Schmid, R. Beeler, S. Giordano, and H. Flinck.
IP and ATM - a position paper.
Technical report, Deliverable ACTS EXPERT; AC094_ATM_21_009.01_CD_NC, (av. at http://lrcwww.epfl.ch/expert/tr17test.ps), 1997.

SCFJ96
H. Schulzrinne, S. Casner, R. Frederick, and V. Jacobson.
RTP: A Transport Protocol for Real-Time Applications.
IETF, RFC 1889, 1996.

Schwartz96
M. Schwartz.
Broadband Integrated Networks.
Prentice Hall PTR, 1996.
ISBN 0-13-519240-4.

Schwieger99
K. Schwieger.
Messung der Charakteristik von TCP Verkehrsströmen in ATM-Netzen.
Studienarbeit, Technische Universität Dresden, Communications Laboratory, Chair for Telecommunications, 1999.

SNH00
N. Seddigh, B. Nandy, and J. Heinanen.
An Assured Rate Per-Domain Behaviour for Differentiated Services.
IETF, draft, draft-ietf-seddigh-pdb-ar-00.txt, (work in progress), 2001.

SNP99
N. Seddigh, B. Nandy, and P. Pieda.
Bandwidth Assurance Issues for TCP flows in a Differentiated Services Network.
Global Telecommunications Conference, GLOBECOM '99, 3:1792-1798, 1999.

SNP99a
N. Seddigh, B. Nandy, and P. Pieda.
Study of TCP and UDP Interaction for th AF PHB.
IETF, draft, draft-nsbnpp-diffserv-tcpudpaf-00.txt, (work in progress), 1999.

SPG97
S. Shenker, C. Partridge, and R. Guerin.
Specification of Guaranteed Quality of Service.
IETF, RFC 2212, 1997.

SW97
S. Shenker and J. Wroclawski.
Network Element Service Specification Template.
IETF, RFC 2216, 1997.

SV95
M. Shreedhar and G. Varghese.
Efficient Fair Queuing using Deficit Round Robin.
Computer Communications Review, 25, 4:231-242, 1995.

Siegel98
E. Siegel.
Rate Control to Enhance Network Service Quality.
Technical report, White Paper, available at www.netreference.com/PublishedArchive/WhitePapers/RateCon_ wp/RateCon.html, 1998.

ST95
K.Y. Siu and H.Y. Tzeng.
Intelligent Congestion Control for ABR Service in ATM Networks.
ACM Computer Communication Review, 5:81-106, 1995.

SH95
O. Spaniol and S. Hoff.
Ereignisorientierte Simulation: Konzepte und Systemrealisierung.
(in german), Thomson Publishing International, 1995.
ISBN 3-8266-0122-X.

SV99
V. Srinivasan and G. Varghese.
Fast Address Lookups Using Controlled Prefix Expansion.
ACM Transactions on Computer Systems, 17,1:1-40, 1999.

Stal98
W. Stallings.
High-Speed Networks: TCP/IP and ATM Design Principles.
Prentice Hall, 1998.
ISBN 0-13-525965-7.

Stevens94
R.W. Stevens.
TCP/IP Illustrated: the protocols, volume 1.
Addison Wesley Longman Inc, 1994.
ISBN 0-201-63346-9.

SXMS00
R. Stewart, Q. Xie, K. Morneault, C. Sharp, et al.
Stream Control Transmission Protocol.
IETF, RFC 2960, 2000.

Sti96
D. Stiliadis.
Traffic Scheduling in Packet-Switched Networks: Analysis, Design, and Implementation.
Ph.D. thesis, University of California, Santa Cruz, USA, 1996.

SV98
D. Stiliadis and A. Varma.
Rate-Proportional Servers: A Design Methodology for Fair Queueing Algorithms.
IEEE/ACM Transactions on Networking, 6, 2:164-174, 1998.

SSZ98
I. Stoica, S. Shenker, and H. Zhang.
Core-Stateless Fair Queueing: Achieving Approximately Fair Bandwidth Allocations in High Speed Networks.
Proceedings of SIGCOMM '98, Vancouver, Canada, pages 118-130, 1998.

SZ99
I. Stoica and H. Zhang.
Providing Guaranteed Services Without Per Flow Management.
Proceedings of SIGCOMM'99, Boston, MA, pages 81-94, 1999.

Sun99
Sun.
Solaris 7 System Administrator Collection: TCP/IP and Data Communications Administration Guide, 1999.

SLSC99
B. Suter, T.V. Lakshman, D. Stiliadis, and A.K. Choudhury.
Buffer Management Schemes for Supporting TCP in Gigabit Routers with Per-Flow Queueing.
IEEE Journal on Selected Areas in Communications (JSAC), 17, 6:1159-1169, 1999.

Swal99
G. Swallow.
MPLS Advantages for Traffic Engineering.
IEEE Communications Magazine, 37, 12:54-57, 1999.

TTL99
L. Taylor, R. Titmuss, and C. Lebre.
The Challenges of Seamless Handover in Future Mobile Multimedia Networks.
IEEE Network, 6, 2:32-37, 1999.

TJL99
M. Tufail, G. Jennes, and G. Leduc.
Attaining per flow QoS with class-based differentiated services.
Proc. of the SPIE '99 conf., Boston, USA, Internet II: Quality of Service and Future Directions, 3842:101-112, 1999.

Turn96
J.S. Turner.
Maintaining high throughput during overload in ATM switches.
Proc. IEEE INFOCOM'96, 1996.

VBJ99
B. Vandalore, G. Babic, and R. Jain.
Analysis and Modeling of Traffic in Modern Data Communication Networks.
submitted to Globecom '99, 1999.

VFJG99b
B. Vandalore, S. Fahmy, R. Jain, R. Goyal, and M. Goyal.
QoS and Multipoint Support for Multimedia Applications over the ATM ABR Service.
IEEE Communications Magazine, 37, 1:53-57, 1999.

VFJG99a
B. Vandalore, S. Fahmy, R. Jain, R. Goyal, and M. Goyal.
General Weighted Fairness and its Support in Explicit Rate Switch Algorithms.
IEEE Computer Communications, 23, 2:149-161, 2000.

BFJF99
B. Vandalore, W.C. Feng, R. Jain, and S. Fahmy.
A Survey of Application Layer Techniques for Adaptive Streaming of Multimedia.
Technical report, OSU-CISRC-5/99-TR14, Ohio State University, Columbus, USA, CIS/OSU, 1999.

VJFD99
B. Vandalore, R. Jain, S. Fahmy, and S. Dixit.
AQuaFWiN: Adaptive QoS Framework for Multimedia in Wireless Networks and.
submitted to LCN '99, 1999.

ATMF970452
B. Vandalore, S. Kalyanaraman, R. Jain, S. Fahmy, et al.
Performance of Bursty World Wide Web (WWW) Sources over ABR.
ATM Forum, document 97-0425, 1997.

VS97
A. Varma and D. Stiliadis.
Hardware Implementation of Fair Queueing Algorithms for Asynchronous Transfer Mode Networks.
IEEE Communications Magazine, 12:54-68, 1997.

Vica97
N. Vicari.
Measurements and Modeling of WWW-Sessions.
Technical report, No. 184, University of Würzburg, Department of Computer Science, 1997.

VKC00
N. Vicari, S. Köhler, and J. Charzinski.
The Dependence of Internet User Traffic Characterization on Access Speed.
Technical report, No. 246, University of Würzburg, Department of Computer Science, 2000.

WML98
J. Waldby, U. Madhow, and T.V. Lakshman.
Total Acknowledgements: A Robust Feedback Mechanism fo End-to-End Congestion Control.
Performance Evaluation Review, Special Issue, SIGMETRICS'98/PERFORMANCE'98, 26, 1:274-275, 1998.

CIFTut97
P. Wang and O. Yaron.
Cells in Frames: The Realization of ATM to the Desktop.
Technical report, available at: http://www.allicancedatacom.com/native-atm-tutorial.htm, 1997.

WLG99
D. Wetherall, U. Legedza, and J. Guttag.
Introducing New Internet Services: Why and How.
IEEE Network, 12, 3:12-19, 1998.

Whet+01
B. Whetten et al.
Reliable Multicast Transport Building Blocks for One-to-Many Bulk-Data Transfer.
IETF, RFC 3048, 2001.

Wroc97
J. Wroclawski.
Specification of the Controlled-Load Network Element Service.
IETF, RFC 2211, 1997.

XN99
X. Xiao and L. Ni.
Internet QoS: A Big Picture.
IEEE Network, 13, 2:8-19, 1999.

YR95
C.Q. Yang and A. Reddy.
A Taxonomy for Congestion Control Algorithms in Packet Switching Networks.
IEEE Network, 07/08:34-45, 1995.

IDYHBB99
R. Yavatkar, D. Hoffman, Y. Bernet, and F. Baker.
SBM (Subnet Bandwidth Manager): A protocol for RSVP-based Admission Control over IEEE 802-style networks, 1999.

IDYPG99
R. Yavatkar, D. Pendarakis, and R. Guerin.
A Framework for Policy-based Admission Control, 1999.

YR99
I. Yeom and A.L.N. Reddy.
Modeling TCP Behavior in a Differentiated Services network.
Technical report, TX 77843-3128, Texas A & M University, College Station, USA, Department of Electrical Engineering, 1999.

YR99c
I. Yeom and N. Reddy.
Impact of marking strategy on aggregated flows in a differentiated-services network.
Proceedings of the International Workshop on QOS (IWQoS), 10 pages, 1999.

YR99a
I. Yeom and N. Reddy.
Realizing throughput guarantees in a differentiated services network.
IEEE International Conference on Multimedia Comp. and Systems, 1999.

YR99b
I. Yeom and N. Reddy.
Marking for QoS Improvement.
accepted for publication in Elsevier Journal of Computer Communications, 2000.

YC99
C. You and K. Chandra.
Time series models for Internet data traffic.
The 24th Annual Conference on Local Computer Networks, LCN '99, pages 164 -171, 1999.

Zha95
H. Zhang.
Service disciplines for guaranteed performance service in packet switching networks.
Proceedings of the IEEE, 83, 10:1374-1399, 1995.

Zha91
L. Zhang.
Virtual Clock: A New traffic control Algorithm for packet switching networks.
ACM Transactions on Computer Systems, 9, 2:101-124, 1991.

OSV94
G. Özveren, R. Simcoe, and G. Varghese.
Reliable and Efficient Hop-by-Hop Flow Control.
ACM SIGCOMM '94, 8, 1994.

About this document ...

Bibliography

This document was generated using the LaTeX2HTML translator Version 99.1 release (March 30, 1999)

Copyright © 1993, 1994, 1995, 1996, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
Copyright © 1997, 1998, 1999, Ross Moore, Mathematics Department, Macquarie University, Sydney.

The command line arguments were:
latex2html -split 0 material

The translation was initiated by Torsten Mueller on 2001-03-06


next_group up previous
Torsten Mueller
2001-03-06