W. Melody Moh

Traffic prediction and dynamic bandwidth allocation over ATM: a neural network approach

ATM (Asynchronous Transfer Mode) has been recommended as the transport vehicle for Broadband Integrated Service Digital Networks (BISDN). The ATM technology offers a great flexibility of transmission bandwidth allocation to accommodate diverse demands of multimedia connections, which include data, voice, video, graphics and images. One major application in ATM networks is to provide real-time, low-loss and minimum-delay transmission of variable bit rate (VBR) traffic which is highly bursty, non-stationary and correlated. In this work we adopt neural network methodology to predict VBR traffic represented by a continuous autoregressive (AR) Markov model. We have found that a simple 1-5-1 backpropagation neural network can accurately predict the VBR traffic. Based on prediction results obtained from neural networks, we propose a dynamic bandwidth allocation scheme for ATM. The proposed scheme is simulated under various traffic loads and buffer sizes. Its performance in terms of cell loss, cell delay and link utilization is examined and compared with two other bandwidth allocation schemes: the static average bandwidth allocation scheme, and the optimal (ideal) bandwidth allocation scheme. Experiments show that for most of the time, performance of the proposed dynamic bandwidth allocation is much better than that of the static average bandwidth allocation, and in many cases it is very close to that of the ideal bandwidth allocation.

Supporting differentiated services with per-class traffic engineering in MPLS

Differentiated services (DiffServ) and MPLS are two major building blocks for providing multi-class services over IP networks. The performance and efficiency of DiffServ architecture can be enhanced with per-class traffic engineering. We propose a new multi-protocol label switching (MPLS) traffic engineering scheme. Based on a constraint-based routing modified from our previously proposed QoS routing algorithm, it enhances the original E-LSP with per-class traffic engineering and load balancing. Major components of the scheme including labeling, load balancing, and routing are described; its features are carefully analyzed against desired traffic engineering requirements. With detailed simulation, we evaluate and compare the proposed scheme with two other schemes: the original E-LSP and E-LSP with load balancing, for their support of expedited forwarding (EF), assured forwarding (AF), and best effort (BE) service classes. We found that, through better utilization of network resources, the proposed scheme is able to accommodate more QoS flows with desired requirements; at the same time it offers better delay and delay jitter for existing EF and AF classes, and improves the overall performance of BE class. We believe that the proposed scheme is an important step for providing scalable, multi-service solution in future IP networks.

Multicasting flow control for hybrid wired/wireless ATM networks

Abstract Multipoint communication has been an increasingly focused topic in computer communication networks, including the Internet, the ATM, and the wireless/mobile networks. The major challenges of designing multicast flow control protocols for a combined wired/wireless network are the varying transmission characteristics (bandwidth, error, reliability, and propagation delay) of the wireless and wired media, and the irregular, different, possibly conflicting flow control requests from multiple receivers in the point-to-multipoint (branching) setting, or from multiple senders in the multipoint-to-point (merging) setting. To address these issues, in this paper we design, analyze, and evaluate both branch-point and merge-point algorithms; we also provide a comparison on the major issues and solutions of the two multicasting scenarios. On point-to-multipoint ABR flow control, we examine an existing max–min fair branch-point algorithm proposed by Siu and Tzeng, and formally analyze its maximum cell loss. A new algorithm is then proposed. Both the maximum cell loss and max–min fairness of new algorithm are analyzed. With extensive simulation, we compare three branch-point algorithms (including a third one proposed by Fahmy, Jain et al.). On multipoint-to-point flow control, we extend the “essential fairness” concept, which was first proposed by Wang and Schwartz to flow control of multicast and unicast TCP traffic in the Internet to the multipoint-to-point ABR flow control. We design a general switch algorithm, which provides essential fairness to the multipoint-to-point ABR flow control. Three major variations of the algorithm, each of them guarantees different fairness (of favor) to unicast or multicast sessions, are then presented. These three schemes are evaluated by simulation. The significance of our approach is illustrated by the formal analysis of cell loss, feedback delay and max–min fairness properties, the generalized fairness definition for multicasting flow control, and the comparisons we made for (1) point-to-multipoint vs. multipoint-to-point scenarios, (2) three branch-point algorithms, and (3) three fairness definitions and mechanisms of multipoint-to-point multicasting. The work may be applied to multicast services in various high speed networks, such as supporting differentiated multicast service over IP, supporting flexible billing schemes, and providing QoS services over hybrid networks.

QoS-guaranteed one-to-many and many-to-many multicast routing

Quality-of-service (QoS) multicast routing with more than one metric has always been technically challenging, since many of them are NP-hard. Most existing QoS multicast routing algorithms are heuristic. Furthermore, many of them considered only the unicast shortest paths, either based on propagation delay or the number of hops. The first part of this paper deals with one-to-many multicast routing. We observed that the end-to-end delay experienced by network packets is determined not only by the propagation distance, but also (and in many cases largely) by the hop-by-hop transmission and queuing delay, which in turn depends on the available bandwidth. Based on this observation, we propose an optimal multicast routing algorithm (the MBDC algorithm), which maximizes the reserved bandwidth while satisfying both delay and bandwidth constraints. Its correctness and time complexity of O(n^2logn) are formally verified, where n is the number of nodes in the network. We also extend the algorithm to support dynamic membership, allowing nodes to join or leave the network, with the cast route updated dynamically. We carefully evaluate both the static and the dynamic algorithms via simulation. The second part of the paper focuses on minimizing the number of centers of multiple-shared multicast trees (MSMT) in many-to-many multicast routing. We propose three heuristic algorithms to this NP-complete problem and compare them with an existing one. We prove, theoretically and through simulation, that our algorithms give the same good results with a significantly shorter running time. We also apply the MBDC extension to the MSMT algorithms, so that the resulting solution not only reduces the number of centers, but also optimizes (maximizes) the bottleneck bandwidth. Proof of correctness, analysis of run-time, and simulation results of these extensions are also provided. We believe that the results report here are significant to the area of QoS-aware multicast routing; the algorithms may be applied to the Internet intra- and inter-domain multicast routing with minimal changes.

An efficient traffic control scheme for integrated voice, video and data over ATM networks: explicit allowed rate algorithm (EARA)

ATM networks are networks with guaranteed quality of service. The main cause of congestion in ATM networks is over utilization of the physical bandwidth. Unlike constant bit rate traffic, the bandwidth reserved by variable bit rate (VBR) traffic is not fully utilized at all instances. Hence, this unused bandwidth is allocated to available bit rate traffic (ABR). As the bandwidth used by VBR traffic changes, the available bandwidth for ABR traffic varies, i.e., the available bandwidth for ABR traffic is inversely proportional to the bandwidth used by the VBR traffic. Based on this fact, a rate based congestion control algorithm, explicit allowed rate algorithm (EARA), is presented. The EARA is compared with the proportional rate control algorithm (PRCA) and explicit rate indication congestion avoidance algorithm (ERICA), under congestion and fairness configurations. The results show that, with a minimal overhead on the switch, the EARA significantly decreases the required buffer space and improves the network throughput.

Correctness and performance of the ATM ABR rate control scheme

Abstract We study both correctness and performance of the source/destination protocols of the available bit rate (ABR) service in asynchronous transfer mode (ATM) networks. Although the basic protocol for rate-based congestion management is relatively simple, the protocol specification has to cope with several “real-world” cases such as failures and delayed/lost feedback which introduce complexity. Rigorous proof of the correct functioning of the protocol based on a formal specification is necessary. We use a formal model to show that the ABR source/destination protocol is free of livelocks, so that under all conditions both resource management (RM) and data cells will be transmitted. Furthermore, if there are data cells available, then the ABR protocol is deadlock free; otherwise, the system goes to a desirable sleep state waiting for data cells, as long as certain parameters are set appropriately at connection setup. We also show that the network options of explicit forward congestion indication (EFCI) and explicit rate (ER) interoperate correctly. In addition to ensuring the correct functioning of the protocol, it is essential that pathological situations do not result in very poor performance, which we view as another form of “incorrect operation”. We derive conditions that ensure that the sources allowed cell rate (ACR) is stable in the presence of delayed or lost feedback RM cells. We arrive at bounds on the number of consecutive RM cell losses tolerated while the ACR rate remains stable. We also provide an asymptotic estimate of ACR and the allowable RM cell loss probability to ensure that ACR is stable, statistically. The ABR protocol contributes to the feedback delay in two ways: the source delay of sending out the probe forward RM (FRM) cells and the destination delay of turning around the backward RM (BRM) cell. We provide a worst-case analysis of the delay in turning around RM cells at the destination station and the worst-case inter-departure time of FRM cells from the source.

Design, analysis, and evaluation of an improved scheme for ATM burst-level admission control

Fast reservation protocols (FRP) have been under consideration for burst-level admission control and block transfer in both ATM LAN and WAN. In this paper we propose a modification to the FRP and the FRP/IT (Immediate Transmission). The resulting protocols, named MFRP (Modified FRP) and MFRP/IT (Modified FRP/IT), are based on bandwidth negotiation between the source and a fast server at the network. Initially, the source requests a desired rate and an acceptable minimum rate. If the network can support at least the minimum rate, it sends an acknowledgment with a granted rate, otherwise a negative acknowledgment is sent. The new mechanism thus uses one extra parameter in both the request and the acknowledgment, but requires no more complicated control mechanism at the network server than the original FRP and FRP/IT protocols. We present mathematical analysis of the new scheme in terms of blocking probability, delay, and wasted bandwidth in unsuccessful multipoint requests. Using computer simulation, the performance of the new protocols are also evaluated and compared with existing protocols: the original FRP, the Adaptive FRP (AFRP), and the FRP/IT. The simulation environment includes both single-stage and multi-stage ATM LAN and WAN for single-rate and multi-rate traffic streams. Simulation results show that the new protocols significantly reduce blocking probability, and, as a result, improve network throughput and message delay; which agrees with what the analysis suggests. Applications of the proposed schemes to block transfer in ATM networks and traffic control in wireless ATM networks are also discussed.