Rate-based Flow Control Research Articles

A Cooperative Differential Game Model for Multiuser Rate-Based Flow Control

Flow control is one typical resource management tool. Its objectives are to adjust the traffic in the network, and resolve data traffic congestion, maximize the utilization of available bandwidth resource and realize fairness among different sources. Taking more consideration on the interaction and the dynamic characteristics of the flow, in this paper, we propose a cooperative differential game of flow control. Our goal is to assign sources with appropriate transmission rate levels so as to the queue length in the bottleneck link is optimal for the network performance and their benefits are maximal. We derive a payment distribution mechanism that would constitute a time-consistent solution and guarantee individual rationality.

Fair resource allocation and stability for communication networks with multipath routing

Multipath networks allow that each source-destination pair can have several different paths for data transmission, thus they improve the performance of increasingly bandwidth-hungry applications and well cater for traffic load balancing and bandwidth usage efficiency. This paper investigates fair resource allocation for users in multipath networks and formulates it as a multipath network utility maximisation problem with several fairness concepts. By applying the Lagrangian method, sub-problems for users and paths are derived from the resource allocation model and interpreted from an economic point of view. In order to solve the model, a novel rate-based flow control algorithm is proposed for achieving optimal resource allocation, which depends only on local information. In the presence of round-trip delays, sufficient conditions are obtained for local stability of the delayed algorithm. As for the end-to-end implementation in Internet, a window-based flow control mechanism is presented since it is more convenient to implement than rate-based flow control.

TCP friendly protocols for media streams over heterogeneous wired–wireless networks

The growing need for Internet friendly streaming protocols prompted us to develop IFTP (Internet Friendly Transport Protocol). IFTP is a protocol with an inherent rate-based flow control mechanism designed to emulate TCP’s window-based flow control mechanism. In this paper, we present two proposals to improve the performance of IFTP in networks with wireless links. The first is IFTP-W, which specifies an error-sensitive section. Bit errors not occurring in this section are ignored. The second is IFTPV, which distinguishes between losses due to congestion and those due to the wireless link. In this paper, we present the performance of IFTP-W with varying error-sensitive section lengths on networks with varying link qualities. We also analyze the performance of IFTPV in extensive scenarios with various network conditions. The results show that both proposals maintain the TCP-friendliness and fairness properties, but perform better than IFTP on wireless networks.

Reinforcement Learning Based Algorithms for Average Cost Markov Decision Processes

This article proposes several two-timescale simulation-based actor-critic algorithms for solution of infinite horizon Markov Decision Processes with finite state-space under the average cost criterion. Two of the algorithms are for the compact (non-discrete) action setting while the rest are for finite-action spaces. On the slower timescale, all the algorithms perform a gradient search over corresponding policy spaces using two different Simultaneous Perturbation Stochastic Approximation (SPSA) gradient estimates. On the faster timescale, the differential cost function corresponding to a given stationary policy is updated and an additional averaging is performed for enhanced performance. A proof of convergence to a locally optimal policy is presented. Next, we discuss a memory efficient implementation that uses a feature-based representation of the state-space and performs TD(0) learning along the faster timescale. The TD(0) algorithm does not follow an on-line sampling of states but is observed to do well on our setting. Numerical experiments on a problem of rate based flow control are presented using the proposed algorithms. We consider here the model of a single bottleneck node in the continuous time queueing framework. We show performance comparisons of our algorithms with the two-timescale actor-critic algorithms of Konda and Borkar (1999) and Bhatnagar and Kumar (2004). Our algorithms exhibit more than an order of magnitude better performance over those of Konda and Borkar (1999).

Smooth switching problem in buffered crossbar switches

Scalability considerations drive the switch fabric design to evolve from output queueing to input queueing and further to combined input and crosspoint queueing (CICQ). However, few CICQ switches are known with guaranteed quality of service, and credit-based flow control induces a scalability bottleneck. In this paper, we propose a novel CICQ switch called the smoothed buffered crossbar or sBUX, based on a new design objective of smoothness and on a new rate-based flow control scheme called the smoothed multiplexer or sMUX. It is proved that with a buffer of just four cells at each crosspoint, sBUX can utilize 100% of the switch capacity to provide deterministic guarantees of bandwidth and fairness, delay and jitter bounds for each flow. In particular, neither credit-based flow control nor speedup is used, and arbitrary fabric-internal latency is allowed between line cards and the switch core.

MAX-MIN FAIR FLOW CONTROL SENSITIVE TO PRIORITIES

Flow control is the dominant technique currently used in communication networks for preventing excess traffic from flooding the network, and for handling congestion. In rate-based flow control, transmission rates of sessions are adjusted in an end-to-end manner through a sequence of operations. In this work, we present a theory of max-min fair, rate-based flow control sensitive to priorities of different sessions, as a significant extension of the classical theory of max-min fair, rate-based flow control to networks supporting applications with diverse requirements on network resources. Each individual session bears a priority function, which maps the session's priority to a transmission rate; the priority is a working abstraction of the session's priority to bandwidth access. Priority functions enable the specification of requirements on bandwidth access by distributed applications, and the formal handling of such requirements. We present priority max-min fairness, as a novel and well motivated fairness condition which requires that assigned rates correspond, through the priority functions, to priorities comprising a max-min vector. We also introduce priority bottleneck algorithms gradually update a session's rate until when its priority is restricted on a priority bottleneck edge of the network. We establish a collection of interesting combinatorial properties of priority bottleneck algorithms. Most significantly, we show that they can only converge to priority max-min fairness. As an application of our general theory, we embed priority bottleneck algorithms in the more realistic optimistic framework for rate-based flow control. The optimistic framework allows for both decreases and increases of session rates. We exploit these additionally provided semantics to prove further combinatorial properties for the termination of priority bottleneck algorithms in the optimistic framework. We use these properties to conclude the first optimistic algorithms for efficient, max-min fair, rate-based flow control sensitive to priorities.

Efficiency of Oblivious versus Nonoblivious Schedulers for Optimistic, Rate-based Flow Control

Two important performance parameters of distributed, rate-based flow control algorithms are their locality and convergence complexity. The former is characterized by the amount of global knowledge that is available to their scheduling mechanisms, while the latter is defined as the number of update operations performed on rates of individual sessions until max-min fairness is reached. Optimistic algorithms allow any session to intermediately receive a rate larger than its max-min fair rate; bottleneck algorithms finalize the rate of a session only if it is restricted by a certain, highly congested link of the network. In this work, we present a comprehensive collection of lower and upper bounds on convergence complexity, under varying degrees of locality, for optimistic, bottleneck, rate-based flow control algorithms. Say that an algorithm is oblivious if its scheduling mechanism uses no information of either the session rates or the network topology. We present a novel, combinatorial construction of a capacitated network, which we use to establish a fundamental lower bound of $\frac{dn}{4} + \frac{n}{2}$ on the convergence complexity of any oblivious algorithm, where n is the number of sessions laid out on a network, and d, the session dependency, is a measure of topological dependencies among sessions. Moreover, we devise a novel simulation proof to establish that, perhaps surprisingly, the lower bound of $\frac{dn}{4} + \frac{n}{2}$ on convergence complexity still holds for any partially oblivious algorithm, in which the scheduling mechanism is allowed to use information about session rates, but is otherwise unaware of network topology. On the positive side, we prove that the lower bounds for oblivious and partially oblivious algorithms are both tight. We do so by presenting optimal oblivious algorithms, which converge after $\frac{dn}{2} + \frac{n}{2}$ update operations are performed in the worst case. To complete the picture, we show that linear convergence complexity can indeed be achieved if information about both session rates and network topology is available to schedulers. We present a counterexample, nonoblivious algorithm, which converges within an optimal number of n update operations. Our results imply a surprising convergence complexity collapse of oblivious and partially oblivious algorithms, and a convergence complexity separation between (partially) oblivious and nonoblivious algorithms for optimistic, bottleneck rate-based flow control.

STABILITY MARGINS FOR A RATE BASED FLOW CONTROL PROBLEM IN MULTIPLE BOTTLENECK NETWORKS

In this paper, the lower bounds for the actual stability margins are derived for a rate–based flow control problem in multiple bottleneck communication networks. Considered stability margins are for the uncertainties in the multiple time–delays and for the rate of change of the time–delays. To observe the change in the actual stability margins with respect to the uncertainties in the time-delays and the rate of change of the time-delays, the lower bounds on the actual stability margins for different cases are depicted.

Centralized Flow Control Scheme for Rate-based Networks

This paper describes the Centralized Flow Control Scheme (CFCS) for a network that uses rate-based flow and congestion control. An example of such a network is an Asynchronous Transfer Mode (ATM) network using the Available Bit Rate (ABR) service. CFCS is a centralized control scheme in which a Network Controller Switch (NCS) maintains the current information of the network such as the number of Virtual Circuits (VCs) active at each switch, available capacity at each switch and rate assigned for each active VC. In order to do this it maintains a Rate Allocation Table (RAT). In our scheme, NCS is the main network controller switch with enough processing power and other switches in the network are the switches without processing power and they do not execute flow control algorithm. We call such switches as Passive Switches (PAS). PAS are controlled by NCS. Thus, in this scheme, the NCS controls the overall operation of the network by allocating rates to ABR connections with the help of RAT. CFCS can achieve zero cell loss, high utilization of links, maintains fairness in allocating bandwidth, it reduces transient response time. Moreover, it needs less feedback information for control and need less parameter to set by users and network administrators. We show the effectiveness of our scheme by simulation experiments.

A Flow Control Scheme Using Broadcast Information for Internet Services in Multiple-Beam Satellite Networks

A new flow control scheme improving performance of the future broadband satellite networks is proposed and analyzed in this paper. The proposed scheme is particularly suitable for the Internet-based services in broadband satellite networks. The complexity of the proposed scheme is comparable with the existing flow control techniques, as it does not require the additional information to be provided by either the satellite or the Earth stations. The throughput and the satellite queue size performances of the proposed scheme are mathematically analyzed and simulated. The results show the significant improvement in the proposed scheme comparing with the conventional window-based and rate-based flow control techniques. Such a scheme can be used in the future satellite networks, which are an important component in the global information infrastructure (GII).

Ping-pong flow control for ATM ABR traffic

In this paper, we propose an improved technique for congestion control, named as ping-pong flow control (PPFC), for asynchronous transfer mode (ATM) available bit rate (ABR) traffic. This is a rate-based flow control scheme, in which the rate regulation is achieved by directly adjusting the transmission rate in the source end station. The proposed algorithm uses a bipolar feedback strategy, which employs positive and negative feedbacks to control the transmission rate for different switch states. These states are determined using the traditional threshold-based method. We also introduce state early detection (SED), which enables the PPFC to control traffic flows more precisely and accurately at critical moments. The simulation results show that the proposed algorithm provides a higher throughput and lower cell loss ratio when compared to the well-known backward explicit congestion notification (BECN). Furthermore, these results also show that PPFC is robust against feedback losses.

Scheduling and Transport for File Transfers on High-Speed Optical Circuits

Scheduling resources on Grids is a well-known problem. The extension of Grids to LambdaGrids requires the scheduling of lambdas, i.e., end-to-end high-speed circuits. In this paper, we propose a scheduling heuristic for such lambdas in support of large-scale scientific applications that require high-throughput transfers of large files. We refer to this heuristic as ‘Varying-Bandwidth List Scheduling’ (VBLS) because the scheduler returns a Time-Range-Capacity (TRC) allocation vector with varying bandwidth levels assigned for different time ranges within the duration of a transfer. The advantage of VBLS over a fixed-bandwidth allocation scheme is that it allows the scheduler to backfill any holes left in resource allocations. Enabling VBLS requires end host applications to specify the file size in their transfer requests. To characterize VBLS, we ran simulation experiments that show that VBLS performance approaches packet-switching performance. This result means that file transfers can take advantage of bandwidth that becomes available subsequent to the start of transfers, a current and critical drawback of typical fixed-bandwidth allocation schemes in circuit-switched networks. Next, we identify the key features needed in a transport protocol that works in conjunction with VBLS and develop the ‘Varying Bandwidth Transport Protocol’ (VBTP). VBTP is a rate based flow control scheme that is coupled with Selective-ARQ based error control. Finally, the paper concludes with a discussion on the impact of transport problems on VBLS scheduling.

A rate-based flow control mechanism for avoiding congestion

The rate-based flow control mechanisms for the Available Bit Rate (ABR) service are used to share the available bandwidth of a bottleneck switch connected to a bottleneck link fairly and reasonably among many competitive users, and to maintain the buffer queue length of the switch at a desired level in order to avoid congestion in Asynchronous Transfer Mode (ATM) networks. In this paper, a control theoretic approach that uses a Deadbeat-Response (DR) controller to the design of a rate-based flow control mechanism is presented. The mechanism has a simple structure and is robust in the sense that its stability is not sensitive to the change of the number of active Virtual Connections (VCs). Simulation results show that this mechanism not only ensures fair share of the bandwidth for all active VCs regardless of the number of hops they traverse but also has the advantages of fast convergence, no oscillation, and high link bandwidth utilization.

A performance prediction model for FECN rate-based flow control in ATM networks: a case study

Forward explicit congestion notification (FECN) plays an important role in rate-based flow control for asynchronous transfer mode (ATM) networks. This paper presents a case study in predicting and evaluating the performance of the FECN flow control. Two network models are under investigation; the first model assumes that a target ATM network has no traffic congestion, while the second model assumes that the network has exhibited heavy congestion. For the light-loaded network, we first estimate the number of cells issued by the source during each time interval and then calculate the required number of resource management (RM) cells in the same time interval. A general analytical form to predict allowable cell rate (ACR) at the source can be finally derived. For the heavy-loaded network, RM cells may be delayed and data cells may be discarded due to severe network congestion. Hence, to derive a general form of ACR, we need to estimate the round-trip time (RTT) of RM cells based on the queuing length of the traversed ATM switches. For the purpose of validation, we perform simulation through OPNET. The performance parameters of our interest include the ACR, the RTT of RM cells, the throughput and the cell loss ratio. We have shown that the differences between the analytical and the simulation results are quite small not only in the ACR adjusting rate, but also in the prediction of the system throughput, the CLR, and the RTT of RM cells.

Design of a new globally stable explicit rate controller for ABR service with saturation

In this paper, we consider the closed-loop rate-based flow control for available bit rate (ABR) service in asynchronous transfer mode (ATM) networks, where the ABR sources adjust their transmission rates in response to feedback explicit rates (ER) from the network switches. The feedback ER may be saturated due to the peak cell rate (PCR) and the minimum cell rate (MCR). The buffer occupancy state of the network switch can also be saturated due to the buffer occupancy. To solve saturation nonlinearity problem and analyze the stability globally, a new ER controller is designed, which is composed of a nonlinear controller and a linear controller. The nonlinear controller is employed to bring buffer occupancy state inside the buffer saturation limits and ensure ER value between PCR and MCR. The linear controller is used to ensure that the buffer occupancy state converges to the desired buffer state. Finally, we demonstrate that good transient and steady-state performances can be achieved through a numerical example.

Comparative study of two flow control mechanisms in high speed networks

Considerable protocol development efforts in recent ATM (Asynchronous Transfer Mode) Forum activities have been focused on the traffic management of available bit rate (ABR) service. It has been shown that ABR service enables persistent, greedy data sources to efficiently utilize ATM network resources with the help of a rate-based flow control mechanism. ATM Forum Traffic Management Specification Version 4.0 document gives a complete description of the end system behavior of the flow control mechanism, but it leaves the details of the switch behavior to be vendor-implementation dependent. For the sake of compatibility and interoperation among flow control mechanisms implemented by vendors, two rate-based mechanisms EPRCA (Enhanced Proportional Rate Control Algorithm) and ERICA (Explicit Rate Indication for Congestion Avoidance) have been recommended in the specification. In this paper, the mechanisms are studied and their performance is analyzed and compared with a material network. Simulation shows that ERICA is significantly better than EPRCA in the performance of steady state and instantaneous state of source end system ACR (Allowed Cell Rate) and buffer queue of bottleneck switch.

Relationship between the buffer threshold and the link utilization in ABR services

Rate-based flow controls play an important role in traffic management of Available Bit Rate (ABR) services. The performance of a rate-based ABR flow control is investigated. In a steady state the evolution of both the Allowed Cell Rate (ACR) and the length of queue within a switch is classified into three cases as the number of active sources varies. In each case, both the link utilization and the maximum queue length are derived. Then, for a given buffer threshold, we determine both the link utilization and the buffer size achieving zero cell loss. Numerical results show the effect of the threshold on the buffer size and the link utilization.

A simplified approach based on source control for ATM ABR service

The available bit rate (ABR) service is designed to provide efficient support of data traffic in ATM networks. A variety of rate-based flow control approaches have been proposed to support ABR service. Most of these approaches improve both the fairness among active connections and the link utilization by putting more and more complexity into a switch. In this article, we propose a flow control approach in which part of rate-calculation work is moved from switches to end-systems. As a result, the proposed approach reduces the rate-calculation effort in the switch. This may reduce the switch complexity and thus the implementation cost. Furthermore, it mitigates the difficulty for setting the measurement interval in switches, as well as features lower and more stable queue occupancy in the network. Simulation results are presented to demonstrate that the proposed approach achieves better performance than ERICA+, a well-known switch algorithm.

Agent-based rate coordination between TCP and ABR congestion control algorithms

In this paper we study the extent of negative interactions between Transmission Control Protocol (TCP) and Asynchronous Transfer Mode (ATM) congestion control through the application of an approach based on coordination via explicit cooperation. To do so, we implement an agent that bridges the gap between the ATM layer and the TCP layer in the protocol stack at the sender end to coordinate the congestion control algorithms of the TCP transport protocol and the ATM cell-oriented switching architecture. The novel idea uses ATM rate-based flow control to adjust the credit-based window size of TCP via the implementation of the agent. The effects of running TCP over Available Bit Rate (ABR) are studied with the help of two simulation models (LAN and WAN). Firstly, we show that TCP window (CWND) and ABR Actual Cell Rate (ACR) are weakly correlated. Secondly, we derive an Agent-based algorithm to ensure that even when the ATM switch queues are indicating no congestion, it is the switch (and not the source) which should control the increment in the TCP source rates. The introduction of the agent at the edge of the network coordinates the TCP and ATM sending transmission and clearly demonstrates a strong correlation coefficient between CWND and ACR. The simulation results show that coordination at the edge of the network makes better use of available bandwidth and improves TCP's performance over ATM for a range of performance metrics such as Cell Loss Ratio, Packet Retransmission Ratio and Throughput.

Optimal structured feedback policies for ABR flow control using two-timescale SPSA

Optimal structured feedback control policies for rate-based flow control of available bit rate service in asynchronous transfer mode networks are obtained in the presence of information and propagation delays, using a numerically efficient two-timescale simultaneous perturbation stochastic approximation (SPSA) algorithm. Models comprising both a single bottleneck node and a network with multiple bottleneck nodes are considered. A convergence analysis of the algorithm is presented. Numerical experiments demonstrate fast convergence even in the presence of significant delays. We also illustrate performance comparisons with the well-known explicit rate indication for congestion avoidance (ERICA) algorithm and describe another algorithm (based on ERICA) that does not require estimating available bandwidth (as in ERICA).