Predictive congestion control of ATM networks: multiple sources/single buffer scenario

Proposes an adaptive methodology using neural networks (NN) for the available bit rate (ABR) service class in a high-speed network. The transmission rates of the sources are controlled in response to the feedback information from the network nodes to prevent congestion. Specifically, the switch and the buffer dynamics are modeled as a nonlinear discrete-time system and a two-layer neural network controller is designed to prevent congestion. Tuning methods are provided for the NN to estimate the unknown traffic. Mathematical analysis is given to demonstrate the stability of the closed-loop error in the buffer occupancy system so that a desired quality of service (QoS) can be guaranteed. The QoS is defined in terms of cell loss ratio, and transmission or transfer delay (latency). No learning phase is required for the NN and initialization of the network weights is straightforward. However, by adding an initial learning phase, the QoS is shown to improve due to low cell losses during transient conditions. Simulation results are provided to justify the theoretical conclusions. Finally, comparison studies are also included to show the effectiveness of the proposed method over adaptive ARMAX, one layer NN and thresholding techniques during simulated congestion.

The broadband integrated services digital networks are based on the ATM technology. As these networks will be deployed gradually, the existing high-speed data services, as Frame Relay and SMDS, will continue to exist and, for interoperability purposes, they will also be supported by the ATM networks. Moreover, ATM networks allows to efficiently support a new service, called available bit rate (ABR) service in the ATM Forum, which requires no bandwidth reservation. When this service is requested, data are transmitted through the network with low priority, being the available network resources shaped dynamically among all the active sources. In this paper, we first discuss how to manage FR and SMDS services in the ATM networks, and report the performance evaluation of the proposed interconnection scenarios; then, we present some preliminary simulation results on the ABR service, whose implementation needs a sophisticated reactive congestion control to obtain an acceptable quality of service.

Performance of TCP traffic and ABR congestion control mechanisms

A rate-based congestion control algorithm regulates cell emission rate of source end systems based on feedback information from the network. It was standardized by the ATM Forum for application to an ABR (available bit rate) service class. In the standard, two types of congestion notification methods of the switch are specified: EFCI marking and explicit-rate marking. In this paper, we focus on the explicit-rate marking switch. We propose our enhancements on a recently proposed switch algorithm known as the max-min scheme. The main objective of our enhancements is to control the queue length of the switch for preventing cell loss and achieving full link-utilization. We show the effectiveness of our switch algorithm by simulation experiments.

The available bit rate (ABR) service category has been approved by the ATM Forum to make use of bandwidth available after satisfying the requirements of real-time services. Thus, many flow control algorithms have been proposed within the context of the ABR service. We have proposed a dynamic rate-based flow control (DRFC) scheme with the primary goal of supporting the ABR service effectively and with the additional advantage of simplicity. This paper investigates the behaviour of the DRFC scheme in conjunction with virtual source/virtual destination (VS/VD) technique. The performance is evaluated in terms of allowed cell rates, number of transmitted cells, buffer occupancy, and link utilization. It has been observed that the implementation of VS/VD technique improves the performance with minimum buffer requirements.

We investigate the performance of the EFCI-based (explicit forward congestion indication) and ER-based (explicit rate; EPRCA in particular) algorithms for the rate-based flow control of the ABR (available bit rate) traffic in an ATM network. We consider the case of multiple switches in tandem. We present several definitions of a bottleneck, and provide conditions that determine which queue is the bottleneck. We show that it is not necessarily the queue with the slowest transmission rate that is the first to notify congestion, and that queues with faster transmission rates can increase considerably the congestion. Moreover, a queue may build up in a faster node before it builds in a slower one. We derive analytic formulas for the maximum queue length. We compare our results to those obtained by approximating a network by a simpler one, containing only the bottleneck switch. We show that the maximum queue lengths under the approximating approach may largely underestimate the ones obtained in the real network.

In this work we examine the problem of regulating available bit rate (ABR) traffic in an ATM network. A unique feature of ABR service is the requirement that sources be regulated using information from distant ATM nodes. This feature is problematic for source control schemes when the time-delays incurred in the feedback path and the temporal variation in the link capacity at the ATM node are considered. A source rate control algorithm that allows for the coordination of the traffic between sources sharing a congested node is presented. The coefficients of the controller are chosen subject to the minimization of a quadratic performance index. The results of the method are also given.

In this paper we present an abstract, rigorous game-theoretic framework for rate-based control of Available Bit Rate (ABR) services in ATM networks. The framework is based on the idea of the Nash arbitration scheme in cooperative game theory which not only provides the rate settings of users which are Pareto optimal from the network point of view of but are also consistent with the fairness axioms of game theory. We first consider the centralized problem and then show that this procedure can be de-centralized such that users can perform the optimization locally such that the overall rates are network optimal. We then consider the problem of charging of ABR connections considering users’ valuations for the best-effort service. We propose a flat-rate charging policy such that the total network revenue is maximum.

We propose a congestion control algorithm and design a state feedback controller for regulating the congestion of available bit rate (ABR) traffic in asynchronous transfer mode (ATM) networks. ABR service is specified as the rate-based feedback mechanism for congestion control and provides a control signal to throttled sources from a bottleneck node in order to adjust the throttled source rates based on available capacity. Our main purposes are as follows: (1) to design a state feedback controller; (2) to place the closed-loop poles in any desired regions based on a pole placement constraint; (3) to use the LMI (linear matrix inequalities) approach for finding the optimal state feedback gain K. The simulation results show that, under pole placement and LMI approaches, our state feedback congestion controller for ABR traffic in ATM networks can possibly guarantee the system stability, provide no buffer overflow, determine the optimal source traffic rates and place the closed-loop poles in any desired regions as well.

This paper presents some studies on the Internet TCP/IP traffic over ATM unspecified bit rate (UBR) and available bit rate (ABR) classes of service. In both cases, fuzzy logic prediction has been used to improve the efficiency and fairness of traffic throughput. For TCP/IP over UBR, a novel fuzzy logic based cell dropping scheme is presented. This is referred to as fuzzy logic selective cell drop (FSCD). A key feature of the scheme is its ability to accept or drop a new incoming packet dynamically based on the predicted future buffer condition in the switch. This is achieved by using fuzzy logic prediction for the production of a drop factor. The packet dropping decision is then based on this drop factor and a predefined threshold value. Simulation results show that the proposed scheme significantly improves the TCP efficiency and fairness. To study TCP/IP over ABR, we applied the fuzzy logic ABR service buffer management scheme from our previous work to both approximate and exact fair rate computation ER switch algorithms. We then compared the performance of the fuzzy logic control with conventional schemes. Simulation results show that on zero TCP packet loss, the fuzzy logic control scheme achieves maximum efficiency and perfect fairness with a smaller buffer size. When mixed with VBR traffic, the fuzzy logic control scheme achieves a higher efficiency with lower cell loss.

This paper presents the Neural Indicates Explicit Rate (NIER) scheme, in which an artificial neural network (ANN) is used to obtain suitable Explicit Rate (ER) for congestion control of Available Bit Rate (ABR) service in ATM networks. The neural network used in this scheme will monitor the queue status in switches’ buffer and output suitable ER value to be carried by Resource Management (RM) cells to the sources for their rate adjustment. A comprehensive comparison study between the proposed algorithm and two well-known algorithms is done and simulation results prove that the proposed algorithm gives good performance in both LAN and WAN topologies.

Acknowledgment bucket scheme for regulating TCP flow over ATM

This paper investigates the impacts of the fast congestion notification function on branch point switch algorithms for multicast (or point-to-multipoint) available bit rate (ABR) flow control in ATM networks. A new backward resource management (BRM) cell-discarding policy is proposed which controls additional BRM cell overhead due to fast congestion notification. The performance of branch-point algorithms with the proposed fast congestion notification is also compared using simulations. The simulation results show that a fast congestion notification is very effective in a severe overload situation, particularly in an initial period with a higher initial cell rate. The fast congestion notification can be also combined with any branch point algorithm. However, its performance is highly dependent on the underlying basic branch point algorithm employed.

High speed networks based on the asynchronous transfer mode (ATM) will use the available bit rate (ABR) service to support best effort type service. The ABR service will use an end-to-end adaptive feedback based rate control, where the sources adapt their rates (at the cell level) based on the state of the network. The performance of this control mechanism depends on a number of control parameters, as well as the delay of the feedback loop. We analyze the feedback based rate control mechanism (as defined by the ATM Forum) and obtain closed-form (approximate) solutions. These design equations will aid in determining appropriate values for the various control parameters.

The key issue at a branch-point switch for point-to-multipoint (multicast) available bit rate (ABR) flow control is how to consolidate backward resource management (BRM) cells returning from each branch to avoid feedback implosion. However, consolidating BRM cells at a branch point can cause undesirable effects such as consolidation noise, consolidation delay, and consolidation loss. This paper first investigates the consolidation problems and proposes their solutions. Furthermore, it introduces various implementation alternatives for the proposed solutions. A combination of the proposed solutions, in which each branch point stores feedback information on a per-branch basis for each multicast virtual connection and only passes BRM cells returning from the farthest destination, provides a good fairness, a higher efficiency and an excellent scalability. In addition, if a fast overload indication is additionally employed, the feedback delay can be significantly reduced in an overload situation.