Showing papers by "Sonia Fahmy published in 1998"
Posted Content•
TL;DR: In this paper, the authors proposed an explicit rate indication scheme for congestion avoidance in ATM networks, where the network switches monitor their load on each link, determining a load factor, the available capacity, and the number of currently active virtual channels.
Abstract: We propose an explicit rate indication scheme for congestion avoidance in ATM networks. In this scheme, the network switches monitor their load on each link, determining a load factor, the available capacity, and the number of currently active virtual channels. This information is used to advise the sources about the rates at which they should transmit. The algorithm is designed to achieve efficiency, fairness, controlled queueing delays, and fast transient response. The algorithm is also robust to measurement errors caused due to variation in ABR demand and capacity. We present performance analysis of the scheme using both analytical arguments and simulation results. The scheme is being implemented by several ATM switch manufacturers.
242 citations
TL;DR: The effects of early packet discard are studied, and a per-VC accounting-based buffer management policy is presented, and the performance of the buffer management policies with various TCP end system congestion control policies are analyzed.
56 citations
29 Mar 1998
TL;DR: Results indicate that the algorithms proposed eliminate the consolidation noise (caused if the feedback is returned before all branches respond), while exhibiting a fast transient response, while the performance of the proposed algorithms is compared under a variety of conditions.
Abstract: ABR traffic management for point-to-multipoint connections controls the source rate to be the minimum rate supported by all the branches of the multicast tree. A number of algorithms have been developed for extending ABR congestion avoidance algorithms to perform feedback consolidation at the branch points. This paper discusses various design options and implementation alternatives for the consolidation algorithms, and proposes a number of new algorithms. The performance of the proposed algorithms and the previous algorithms is compared under a variety of conditions. Results indicate that the algorithms we propose eliminate the consolidation noise (caused if the feedback is returned before all branches respond), while exhibiting a fast transient response.
52 citations
TL;DR: It is shown that with a good switch algorithm, ABR pushes congestion to the edges of the ATM network while UBR leaves it inside the ATM portion, so the switch ABR buffer requirement for zero-packet-loss high-throughput TCP transmission is a sublinear function of the number of TCP connections.
Abstract: ATM networks are quickly being adopted as backbones over various parts of the Internet. This article studies the dynamics and performance of the TCP/IP protocol over the ABR and UBR services of ATM networks. Specifically the buffering requirements in the ATM switches as well as the ATM edge devices. It is shown that with a good switch algorithm, ABR pushes congestion to the edges of the ATM network while UBR leaves it inside the ATM portion. As a result, the switch ABR buffer requirement for zero-packet-loss high-throughput TCP transmission is a sublinear function of the number of TCP connections.
44 citations
13 Oct 1998
TL;DR: A general definition of weighted fairness is given and how a pricing policy can be mapped to general weighted (GW) fairness is discussed to demonstrate that, the modified switch algorithm achieves GW fairness.
Abstract: We give a general definition of weighted fairness and discuss how a pricing policy can be mapped to general weighted (GW) fairness. The GW fairness can be achieved by calculating the excess fairshare (weighted fairshare of the left over bandwidth) for each VC. We show how a switch algorithm can be modified to support the GW fairness by using the excess fairshare term. We use ERICA+ as an example switch algorithm and show how it can be modified to achieve the general fairness. Simulations results are presented to demonstrate that, the modified switch algorithm achieves GW fairness. An analytical proof for convergence of the modified ERICA+ algorithm is given in the appendix.
34 citations
07 Jun 1998
TL;DR: This work proposes a method for determining the number of active connections and the fair bandwidth share for each, and proves the efficiency and fairness of the proposed method analytically, and simulate it for a number of configurations.
Abstract: The available bit rate (ABR) service is designed to fairly allocate the bandwidth unused by higher priority services. The network indicates to the ABR sources the rates at which they should transmit to minimize their cell loss. Switches must constantly measure the demand and available capacity, and divide the capacity fairly among the contending connections. In order to compute the fair and efficient allocation for each connection, a switch needs to determine the effective number of active connections. We propose a method for determining the number of active connections and the fair bandwidth share for each. We prove the efficiency and fairness of the proposed method analytically, and simulate it for a number of configurations.
30 citations
Posted Content•
TL;DR: This work studies the performance of selective acknowledgements (SACK) with TCP over the ATM-UBR service category and examines various unspecified bit rate (UBR) drop policies, TCP mechanisms and network configurations to recommend optimal parameters for TCP over UBR.
Abstract: We study the performance of Selective Acknowledgments with TCP over the ATM-UBR service category. We examine various UBR drop policies, TCP mechanisms and network configurations to recommend optimal parameters for TCP over UBR. We discuss various TCP congestion control mechanisms compare their performance for LAN and WAN networks. We describe the effect of satellite delays on TCP performance over UBR and present simulation results for LAN, WAN and satellite networks. SACK TCP improves the performance of TCP over UBR, especially for large delay networks. Intelligent drop policies at the switches are an important factor for good performance in local area networks.
24 citations
Posted Content•
TL;DR: The inverse hyperbolic is the best queue control function to reduce complexity, and the linear function can be used since it performs satisfactorily in most cases.
Abstract: The main goals of a switch scheme are high utilization, low queuing delay and fairness. To achieve high utilization the switch scheme can maintain non-zero (small) queues in steady state which can be used if the sources do not have data to send. Queue length (delay) can be controlled if part of the link capacity is used for draining queues in the event of queue build up. In most schemes a simple threshold function is used for queue control. Better control of the queue and hence delay can be achieved by using sophisticated queue control functions. It is very important to design and analyze such queue control functions. We study step, linear, hyperbolic and inverse hyperbolic queue control functions. Analytical explanation and simulation results consistent with analysis are presented. From the study, we conclude that inverse hyperbolic is the best control function and to reduce complexity the linear control function can be used since it performs satisfactorily in most cases.
24 citations
Posted Content•
TL;DR: A model to study the end-to-end delay performance of a satellite-ATM netowrk and a simulation model to calculate the buffering delay for TCP/IP traffic over ATM ABR and UBR service categories are presented.
Abstract: In this paper we present a model to study the end-to-end delay performance of a satellite-ATM netowrk. We describe a satellite-ATM network architecture. The architecture presents a trade-off between the on-board switching/processing features and the complexity of the satellite communication systems. The end-to-end delay of a connection passing through a satellite constellation consists of the transmission delay, the uplink and downlink ground terminal-satellite propagation delay, the inter-satellite link delays, the on-board switching, processing and buffering delays. In a broadband satellite network, the propagation and the buffering delays have the most impact on the overall delay. We present an analysis of the propagation and buffering delay components for GEO and LEO systems. We model LEO constellations as satellites evenly spaced in circular orbits around the earth. A simple routing algorithm for LEO systems calculates locally optimal paths for the end-to-end connection. This is used to calculate the end-to-end propagation delays for LEO networks. We present a simulation model to calculate the buffering delay for TCP/IP traffic over ATM ABR and UBR service categories. We apply this model to calculate total end-to-end delays for TCP/IP over satellite-ATM networks.
23 citations
TL;DR: This paper model a VS/VD ATM switch and study the issues in designing the coupling between ABR segments, identifying a number of implementation options for the coupling and showing that the choice of the implementation option significantly affects the system performance.
Abstract: The Available Bit Rate (ABR) service in ATM networks uses end-to-end rate-based flow control to allow fair and efficient support of data applications over ATM networks. One of the architectural features in the ABR specification [ATM Forum, ATM Traffic Management Specification Version 4.0, April 1996] is the Virtual Source/Virtual Destination (VS/VD) option. This option allows a switch to divide an end-to-end ABR connection into separately controlled ABR segments by acting like a (virtual) destination on one segment, and like a (virtual) source on the other. The translation and propagation of feedback in the VS/VD switch between the two ABR control segments (called “coupling”) is implementation specific. In this paper, we model a VS/VD ATM switch and study the issues in designing the coupling between ABR segments. We identify a number of implementation options for the coupling and show that the choice of the implementation option significantly affects the system performance in terms of (a) the system stability in the steady state, (b) the time to respond to transient changes and converge to the steady state, and (c) the buffer requirements at the switches.
20 citations
07 Oct 1998
TL;DR: In this paper, the authors give various possibilities for defining fairness for multipoint connections, and show the tradeoffs involved, and also show that ATM bandwidth allocation algorithms need to be adapted to give fair allocations for multippoint-to-point connections.
Abstract: In multipoint-to-point connections, the traffic at the root is the combination of all traffic originating at the leaves. A crucial concern in the case of multiple senders is how to define fairness within a multicast group, and among groups and point-to-point connections. Fairness definition can be complicated since the multipoint connection can have the same identifier on each link, and senders might not be distinguishable in this case. Many rate allocation algorithms implicitly assume that there is only one sender in each VC, which does not hold for multipoint-to-point cases. We give various possibilities for defining fairness for multipoint connections, and show the tradeoffs involved. In addition, we show that ATM bandwidth allocation algorithms need to be adapted to give fair allocations for multipoint-to-point connections.© (1998) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
26 May 1998
TL;DR: It is concluded that when the entire link capacity is allocate to GFR traffic, per-VC scheduling is necessary to provide minimum throughput guarantees to TCP traffic.
Abstract: We describe proposed enhancements to the ATM unspecified bit rate (UBR) service that guarantee a minimum rate at the frame level to the UBR VCs. These enhancements have been called guaranteed frame rate (GFR). In this paper, we discuss the motivation, design and implementation issues for GFR. We present the design of buffer management and tagging mechanisms to implement GFR. We study the effects of policing, per-VC buffer allocation, and per-VC queueing on providing GFR to TCP/IP traffic. We conclude that when the entire link capacity is allocate to GFR traffic, per-VC scheduling is necessary to provide minimum throughput guarantees to TCP traffic. We examine the role of frame tagging in the presence of scheduling and buffer management for providing minimum rate guarantees.
TL;DR: In this article, the performance of TCP/IP protocols over ATM network's Available Bit Rate and Unspecified Bit Rate (UBR) services is analyzed. And it is shown that ABR pushes congestion to the edges of the ATM network while UBR leaves it inside the ATM portion.
Abstract: The Asynchronous Transfer Mode (ATM) networks are quickly being adopted as backbones over various parts of the Internet. This paper analyzes the performance of TCP/IP protocols over ATM network's Available Bit Rate (ABR) and Unspecified Bit Rate (UBR) services. It is shown that ABR pushes congestion to the edges of the ATM network while UBR leaves it inside the ATM portion.
12 Oct 1998
TL;DR: In this article, the authors study the design and analysis of several queue control functions: the step, linear, hyperbolic and inverse Hyperbolic functions, and they conclude that the inverse HHB is the best queue control function.
Abstract: The ABR rate allocation schemes can achieve high link utilizations by maintaining non-zero (small) queues in the steady state, and draining queues when the sources do not have data to send. The queue length (and queuing delays) can be controlled if part of the available bandwidth is used for draining queues in the event of queue build up. A simple threshold function can allocate such bandwidth to drain queues. Better control of the queues, and hence delay, can be achieved using more sophisticated queue control functions. We study the design and analysis of several such queue control functions: the step, linear, hyperbolic and inverse hyperbolic functions. Analytical explanation and simulation results consistent with analysis are presented. From the study, we conclude that the inverse hyperbolic is the best queue control function. To reduce complexity, the linear function can be used since it performs satisfactorily in most cases.
11 Oct 1998
TL;DR: This work studies the effect of VBR carrying long-range dependent, multiplexed MPEG-2 video sources traffic on ABR carrying TCP traffic, and finds that a switch algorithm like ERICA+ can tolerate this variance in ABR capacity while maintaining high throughput and low delay.
Abstract: Compressed video is well known to be self-similar in nature. We model VBR carrying long-range dependent, multiplexed MPEG-2 video sources traffic. The actual traffic for the model is generated using fast-Fourier transform of the fractional Gaussian noise sequence. Our model of compressed video sources bears similarity to an MPEG-2 transport stream carrying video, i.e., it is long-range dependent and generates traffic in a piecewise constant bit rate manner. We study, the effect of such VBR traffic on ABR carrying TCP traffic. The effect of such VBR traffic is that the ABR capacity is highly variant. We find that a switch algorithm like ERICA+ can tolerate this variance in ABR capacity while maintaining high throughput and low delay. We present simulation results for terrestrial and satellite configurations.
11 Oct 1998
TL;DR: It is shown that under conditions of low buffer allocation, it is possible to control TCP rates with FIFO queuing and buffer management, and a buffer management policy is presented that provides loose rate guarantees to SACK TCP sources when the total buffer allocation is low.
Abstract: The ATM guaranteed frame rate (GFR) service is intended for best effort traffic that can benefit from minimum throughput guarantees. Edge devices connecting LANs to an ATM network can use GFR to transport multiple TCP/IP connections over a single GFR VC. These devices would typically multiplex VCs into a single FIFO queue. It has been shown that in general, FIFO queuing is not sufficient to provide rate guarantees, and per-VC queuing with scheduling is needed. We show that under conditions of low buffer allocation, it is possible to control TCP rates with FIFO queuing and buffer management. We present analysis and simulation results on controlling TCP rates by buffer management. We present a buffer management policy that provides loose rate guarantees to SACK TCP sources when the total buffer allocation is low. We study the performance of this buffer management scheme by simulation.
Posted Content•
TL;DR: With the ERICA [erica-final] switch algorithm, the buffering required is independent of the number of TCP sources, and the maximum buffers required at the switch is proportional to the maximum round trip time (RTT) of all VCs through the link.
Abstract: We study the buffering requirements for zero cell loss for TCP over ABR. We show that the maximum buffers required at the switch is proportional to the maximum round trip time (RTT) of all VCs through the link. The number of round-trips depends upon the the switch algorithm used. With our ERICA [erica-final] switch algorithm, we find that the buffering required is independent of the number of TCP sources. We substantiate our arguments with simulation results.
Posted Content•
TL;DR: In this article, the design issues in improving TCP performance over the ATM UBR service are studied and various design options available to the network as well as to the end systems to improve TCP performance.
Abstract: In this paper we study the design issues in improving TCP performance over the ATM UBR service. ATM-UBR switches respond to congestion by dropping cells when their buffers become full. TCP connections running over UBR can experience low throughput and high unfairness. Intelligent switch drop policies and end-system policies can improve the performance of TCP over UBR with limited buffers. We describe the various design options available to the network as well as to the end systems to improve TCP performance over UBR. We study the effects of Early Packet Discard, and two per-VC accounting based buffer management policies. We also study the effects of various TCP end system congestion control policies including slow start and congestion avoidance, fast retransmit and recovery and selective acknowledgments. We present simulation results for various small and large latency configurations with varying buffer sizes and number of sources.
Posted Content•
TL;DR: The analysis presented here led to the changes which are now part of the ATM traffic management (TM 4.0) specification, where the size of the transient buffer exposure (TBE) parameter was set to 24 bits, and no size was enforced for the Crm parameter.
Abstract: During the design of ABR traffic management at the ATM Forum, we performed several analyses to ensure that the ABR service will operate efficiently over satellite links. In the cases where the performance was unacceptable, we suggested modifications to the traffic management specifications. This paper describes one such issue related to the count of missing resource management cells (Crm) parameter of the ABR source behavior. The analysis presented here led to the changes which are now part of the ATM traffic management (TM 4.0) specification. In particular, the size of the transient buffer exposure (TBE) parameter was set to 24 bits, and no size was enforced for the Crm parameter. This simple change improved the throughput over OC-3 satellite links from 45 Mbps to 140 Mbps.
01 Jan 1998
Posted Content•
TL;DR: This paper discusses the design, development, and the final shape of these policies and their impact on the ABR service, and compares the various alternatives through a performance evaluation.
Abstract: The Available Bit Rate (ABR) service has been developed to support 21st century data applications over Asynchronous Transfer Mode (ATM). The ABR service uses a closed-loop rate-based traffic management framework where the network divides left-over bandwidth among contending sources. The ATM Forum traffic management group also incorporated open-loop control capabilities to make the ABR service robust to temporary network failures and source inactivity. An important problem addressed was whether rate allocations of sources should be taken away if sources do not use them. The proposed solutions, popularly known as the Use-It-or-Lose-It (UILI) policies, have had significant impact on the ABR service capabilities. In this paper we discuss the design, development, and the final shape of these policies and their impact on the ABR service. We compare the various alternatives through a performance evaluation.
TL;DR: It is shown that ATM bandwidth allocation algorithms need to be adapted to give fair allocations for multipoint-to-point connections, and the tradeoffs involved.
Abstract: In multipoint-to-point connections, the traffic at the root (destination) is the combination of all traffic originating at the leaves. A crucial concern in the case of multiple senders is how to define fairness within a multicast group and among groups and point-to-point connections. Fairness definition can be complicated since the multipoint connection can have the same identifier (VPI/VCI) on each link, and senders might not be distinguishable in this case. Many rate allocation algorithms implicitly assume that there is only one sender in each VC, which does not hold for multipoint-to-point cases. We give various possibilities for defining fairness for multipoint connections, and show the tradeoffs involved. In addition, we show that ATM bandwidth allocation algorithms need to be adapted to give fair allocations for multipoint-to-point connections.
Posted Content•
TL;DR: The OPNET models that have been developed for ATM and ABR design and analysis will be described, which allow users to integrate voice, video, and data on the same network.
Abstract: Asynchronous transfer mode (ATM) is the new generation of computer and communication networks that are being deployed throughout the telecommunication industry as well as in campus backbones. ATM technology distinguishes itself from the previous networking protocols in that it has the latest traffic management technology and thus allows guaranteeing delay, throughput, and other performance measures. This in turn, allows users to integrate voice, video, and data on the same network. Available bit rate (ABR) service in ATM has been designed to fairly distribute all unused capacity to data traffic and is specified in the ATM Forum's Traffic Management (TM4.0) standard. This paper will describe the OPNET models that have been developed for ATM and ABR design and analysis.
TL;DR: In this article, the authors discuss the design, development, and the final shape of these policies and their impact on the Available Bit Rate (ABR) service and compare the various alternatives through a performance evaluation.
Abstract: The Available Bit Rate (ABR) service has been developed to support data applications over Asynchronous Transfer Mode (ATM) The ABR service uses a closed-loop rate-based traffic management framework where the network divides left-over bandwidth among contending sources The ATM Forum traffic management group also incorporated open-loop control capabilities to make the ABR service robust to temporary network failures and source inactivity An important problem addressed was whether rate allocations of sources should be taken away if sources do not use them The proposed solutions, popularly known as the Use-It-or-Lose-It (UILI) policies, have had significant impact on the ABR service capabilities In this paper we discuss the design, development, and the final shape of these policies and their impact on the ABR service We compare the various alternatives through a performance evaluation
TL;DR: In this article, the authors show that ABR scales well with a large number of bursty TCP sources in the system, which raises the possibility of large switch queues if the source rates are not controlled by ABR.
Abstract: We model World Wide Web (WWW) servers and clients running over an ATM network using the ABR (available bit rate) service. The WWW servers are modeled using a variant of the SPECweb96 benchmark, while the WWW clients are based on a model by Mah. The traffic generated by this application is typically bursty, i.e., it has active and idle periods in transmission. A timeout occurs after given amount of idle period. During idle period the underlying TCP congestion windows remain open until a timeout expires. These open windows may be used to send data in a burst when the application becomes active again. This raises the possibility of large switch queues if the source rates are not controlled by ABR. We study this problem and show that ABR scales well with a large number of bursty TCP sources in the system.
Posted Content•
TL;DR: The problems in designing satellite network protocols, as well as some of the solutions proposed to tackle these problems, will be the primary focus of this survey.
Abstract: Satellite communication systems are the means of realizing a global broadband integrated services digital network. Due to the statistical nature of the integrated services traffic, the resulting rate fluctuations and burstiness render congestion control a complicated, yet indispensable function. The long propagation delay of the earth-satellite link further imposes severe demands and constraints on the congestion control schemes, as well as the media access control techniques and retransmission protocols that can be employed in a satellite network. The problems in designing satellite network protocols, as well as some of the solutions proposed to tackle these problems, will be the primary focus of this survey.
08 Nov 1998
TL;DR: VS/VD can help in limiting buffer requirements of switches, based on the length of their VS/VD control loops, and is especially useful in isolating terrestrial networks from the effects of long delay satellite networks by limiting thebuffer requirements of the terrestrial switches.
Abstract: We describe issues in designing rate allocation schemes in ATM-ABR networks for virtual source/virtual destination (VS/VD) switches. We propose a rate allocation scheme for VS/VD switches that uses per-VC queuing and per-VC control. We analyze the performance of this scheme, and conclude that VS/VD can help in limiting buffer requirements of switches, based on the length of their VS/VD control loops. VS/VD is especially useful in isolating terrestrial networks from the effects of long delay satellite networks by limiting the buffer requirements of the terrestrial switches.
01 Jan 1998
07 Oct 1998
TL;DR: This work model WWW servers and clients running over an ATM network using the available bit rate (ABR) service and shows that ABR scales well to a large number of bursty TCP sources in the system.
Abstract: Asynchronous transfer-mode (ATM) is the technology chosen for implementing the broadband integrated services digital network. The performance of internet protocols over ATM is an extremely important research area. As web traffic forms a major portion of the Internet traffic, we model WWW servers and clients running over an ATM network using the available bit rate (ABR) service. The WWW servers are modeled using a variant of the SPECweb96 benchmark, while the WWW clients are based on a model proposed. The traffic generated is typically bursty, having active and idle transmission periods. A timeout occurs after a certain idle interval. During idle periods, the underlying TCP congestion windows remain large until the timer expires. This raises the possibility of large queues at the switches, if the source rates ar not controlled by ABR. We study this problem and show that ABR scales well to a large number of bursty TCP sources in the system.© (1998) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
Posted Content•
TL;DR: In this article, the authors study the effect of VBR traffic on ABR carrying TCP traffic and find that the ABR capacity is highly variant, and they find that a switch algorithm like ERICA+ can tolerate this variance in ABR while maintaining high throughput and low delay.
Abstract: Compressed video is well known to be self-similar in nature. We model VBR carrying Long-Range Dependent (LRD), multiplexed MPEG-2 video sources. The actual traffic for the model is generated using fast-fourier transform of generate the fractional gaussian noise (FGN) sequence. Our model of compressed video sources bears similarity to an MPEG-2 Transport Stream carrying video, i.e., it is long-range dependent and generates traffic in a piecewise-CBR fashion. We study the effect of such VBR traffic on ABR carrying TCP traffic. The effect of such VBR traffic is that the ABR capacity is highly variant. We find that a switch algorithm like ERICA+ can tolerate this variance in ABR capacity while maintaining high throughput and low delay. We present simulation results for terrestrial and satellite configurations.