This paper presents Random Early Detection (RED) gate-ways for congestion avoidance in packet-switched networks. The gateway detects incipient congestion by com-puting the average queue size. The gateway could notify connections of congestion either by dropping packets ar-riving at the gateway or by setting a bit in packet headers. When the average queue size exceeds a preset threshold,the gateway drops or marks each arriving packet with a certain probability, where the exact probability is a func-tion of the average queue size. RED gateways keep the average queue size low while allowing occasional bursts of packets in the queue. During congestion, the probability that the gateway notifies a particular connection to reduce its window is roughly proportional to that connection's share of the bandwidth throughthe gateway. RED gateways are designed to accompany a transport-layer congestion control protocol such as TCP.The RED gateway has no bias against bursty traffic and avoids the global synchronization of many connectionsdecreasing their window at the same time. Simulations of a TCP/IP network are used to illustrate the performance of RED gateways.

This paper examines the performance of TCP/IP, the Internet data transport protocol, over Wide Area Networks (WANs) in which data traffic could coexist with real-time traffic such as voice and video. Specifically, we attempt to develop a basic understanding, using analysis and simulation, of the properties of TCP/IP in a regime where (1) the bandwidth-delay product of the network is high compared to the buffering in the network, and (2) there may be transient congestion due to fluctuations in real-time traffic, modeled here as producing random losses among the packets of the TCP connection of interest. The following key results are obtained. First, random loss leads to significant throughput deterioration when the product of the loss probability and the square of the bandwidth-delay product is larger than one. Unless network resources are specifically reserved for data traffic, data traffic will inevitably incur random losses due to transient fluctuations in higher priority real-time traffic when the network is highly utilized. Second, for multiple connections sharing a bottleneck link, TCP is grossly unfair towards connections with higher round-trip delays. This means that a simple First In First Out (FIFO) queueing discipline might not suffice for data traffic in WANs. Finally, we observe that, while the recent Reno version of TCP produces less bursty traffic than the original Tahoe version, it is less robust than the latter when successive losses are closely spaced. We conclude by indicating modifications that may be required both at the transport and network layers to provide good end-to-end performance over high-speed WANs.

This paper presents a control-theoretic approach to reactive flow control in networks that do not reserve bandwidth. We assume a round-robin-like queue service discipline in the output queues of the network’s switches, and propose deterministic and stochastic models for a single conversation in a network of such switches. These models motivate the Packet-Pair rate probing technique, and a provably stable rate-based flow control scheme. A Kalman state estimator is derived from discrete-time state space analysis, but there are difficulties in using the estimator in practice. These difficulties are overcome by a novel estimation scheme based on fuzzy logic. We then present a technique to extract and use additional information horn the system to develop a continuous-time system model. This is used to design a wuisnt of the control law that is also provably stable, and, in addition, takes control action as rapidly as possible. Finally, practical issues such as correcting parameter drift and cmmlination with window flow control are described.

this paper, we will assume that all window sizes are measured in units of maximum size packets, instead of bytes. In the original TCP specification [10], the window used by the sender, which we will denote by wnd, is the 1

In this paper we explore the bias in TCP/IP networks against connections with multiple congested gateways. We consider the interaction between the bias against connections with multiple congested gateways, the bias of the TCP window modification algorithm against connections with longer roundtrip times, and the bias of Drop Tail and Random Drop gateways against bursty traffic. Using simulations and a heuristic analysis, we show that in a network with the window modification algorithm in 4.3 tahoe BSD TCP and with Random Drop or Drop Tail gateways, a longer connection with multiple congested gateways can receive unacceptably low throughput. We show that in a network with no bias against connections with longer roundtrip times and with no bias against bursty traffic, a connection with multiple congested gateways can receive an acceptable level of throughput. We discuss the application of several current measures of fairness to networks with multiple congested gateways, and show that diff...

This paper studies a simple strategy, proposed independently by Gallager [1] and Katevenis [2], for fairly allocating link capacity in a point-to-point packet network with virtual circuit routing. Each link offers its packet transmission slots to its user sessions by polling them in round-robin order. In addition, window flow control is used to prevent excessive packet queues at the network nodes. As the window size increases, the session throughput rates are shown to approach limits that are perfectly fair in the max-min sense. That is, the smallest session rate in the network is as large as possible and, subject to that constraint, the second-smallest session rate is as large as possible, etc. If each session has periodic input (perhaps with jitter) or has such heavy demand that packets are always waiting to enter the network, then a finite window size suffices to produce perfectly fair throughput rates. The round-robin method is considerably simpler than earlier strategies for achieving global fairness. The fair session rates are not explicitly computed, and the only overhead communication is that required for the window acknowledgments. The main drawback is that large windows are needed to achieve even approximately fair throughputs in some (hopefully rare) situations, and large windows permit large cross-network delays. Fortunately, the round-robin method offers other throughput guarantees that, while falling short of perfect fairness, do apply even for sessions with small windows. Such sessions are promised reasonable bounds on their cross-network packet delay as well.

Dynamics of Internet load are investigated using statistics of round-trip delays, packet losses and out-of-order sequence of acknowledgments. Several segments of the Internet are studied. They include a regional network (the Jon yon Neumann Center Network), a segment of the NSFNet backbone and a cross-country network consisting of regional and backbone segments.

This paper presents the packet-pair rate-based feedback flow control scheme. This scheme is designed for networks where individual connections do not reserve bandwidth and for the available bitrate (best-effort) component of integrated networks. We assume a round-robin-like queue service discipline in the output queues of the network's switches, and propose a linear stochastic model for a single conversation in a network of such switches. These model motivates the Packet-Pair rate probing technique, which forms the basis for provably stable discrete and continuous time rate-based flow control schemes. We present a novel state estimation scheme based on fuzzy logic. We then address several practical concerns: dealing with system startup, retransmission and timeout strategy, and dynamic setpoint probing. We present a finite state machine as well as source code for a model implementation. The dynamics of a single source, the interactions of multiple sources, and the behavior of packet-pai...

Weaknesses in several recently proposed ideas about congestion control and avoidance in high-speed networks are identified. Both sides of the debate concerning priorreservation of resources versus walk-in service, open-loop control versus feedback control, rate control versus window control, and router-based control versus source-based control are presented. The circumstances under which backpressure is useful or not useful are discussed, and it is argued that a single congestion scheme is not sufficient, but that a combination of several schemes is required for complete congestion management in a network. Key Words: Congestion Control, Flow Control, Resource Management, Overload Control, Congestion Avoidance, High-Speed Networks. Introduction With the advent of gigabits per second links and networks, the interest in congestion management 2 is increasing. This is because high-speed links have to coexist with lowspeed networks of the past that will continue to be used for quite some...

This paper gives a class of flow control algorithms for the adaptive allocation of bandwidths to virtual connections (VC) in high speed, wide area ATM networks. The feedback rate to the source from the network is parsimonious, with each feedback bit indicating whether the buffer at a distant switch is above or below a threshold. The service discipline at the switch is First-Come-First-Served. The important goal of adaptability aims to make all of the network bandwidth available to the active VCs, even though the number of such VCs is variable over a given range. Each VC has two parameters, one giving its minimum guaranteed bandwidth and the other is the weight for determining its share of the uncommitted bandwidth. Judicious selection of these parameters defines distinctive services, such as Best Effort and Best Effort with Minimum Bandwidth. We derive design rules for selecting the parameters of the algorithms such that the appropriate guarantees and fairness properties are exhibited ...