Asynchronous transfer mode

This thesis addresses studies on cost-functions developed on the basis of maximum entropy principle, for applications in artificial neural network (ANN) optimization endeavors. The maximization of entropy refers to maximizing Shannon information pertinent to the difference in the output and the teacher value of an ANN. Apart from the Shannon format of the negative entropy formulation a set of Csiszar family functions are also considered. The error-measures obtained, via these maximum entropy formulations are adopted as cost-functions in the training and prediction schedules of a test perceptron. A comparative study is done on the performance of these cost-functions in facilitating the test network towards optimization so as to predict a standard teacher function sin (.). The study is also extended to predict a parameter (such as cell delay variation) in a practical ATM telecommunication system. Concluding remarks and scope for an extended study are also indicated.

ATM multiplexing, (which is essentially a statistical multiplexing process with priority constraints) leads to a departure pattern of cells on the trunk (or bearer) line, characterized by the profiles of incoming traffics at the multiplexer. Such profiles are decided by the bit-rate variations and workload fluctuations of the associated sources, namely voice, data and video of the cells. Bit rate statistics of incoming traffic with priority constraints decide the clustered and/or non-clustered epochs of cells on the multiplexed fine. If the incoming traffic is delay-sensitive, it gets the priority in the multiplexing strategy and the associated cells inbiting no clusters; whereas, the non-isochronous traffics (facing low priority flow service) induce stagnation in the queue and pose eventually a clustered structure of cells on the trunk line. Workload fluctuation is largely application-dependent. The corresponding teletraffic statistics would also show deviations in the performance from the mean-field regime. In essence, the cell-stream departing from ATM multiplexer would exhibit statistics of deviations. In the present study, the associated delay parameter is considered and modeled under constant and variable bit rate conditions for fluctuating workloads. Mean-field theory is applied in modeling constant bit rate traffics. For variable and bursty rate transmissions, it is shown that the delay is governed by a non-homogeneous compound Poisson process. Relevant statistical bounds are established via large-deviation theory. Determination of the bounds of ATM cell-delay variations is important in developing call admission control algorithms.

High speed ATM networks support a variety of communication services, that have different traffic characteristics, which causes the network to be congested quickly. An ATM network with different communication services, data, voice and video, is simulated to study the effect of congestion on network operation. A modified leaky bucket mechanism is used to shape the traffic entering the network, which improved the performance in terms of cell losses and cell delay. The original leaky bucket mechanism is so conservative, that it drops a large number of ATM cells. Another scheme called virtual leaky bucket is proposed in this thesis. In this scheme violating cells are marked and then allowed to enter the network. The scheme is simulated and its performance is compared to the leaky bucket mechanism. Shaped virtual leaky bucket with marking is shown to have much better performance as long as the minimum requirements of non-violating cells are guaranteed.

Asynchronous Transfer Mode (ATM) Networks are based on connections that are multiplexed dynamically on a single link. For satisfactory performance in an ATM network, congestion control is extremely necessary to guarantee the negotiated Quality of Service (QoS) for every connection. Traffic shaping is a congestion control mechanism that alters the traffic characteristics of a stream of cells on a connection to achieve better network efficiency by meeting the QoS objectives. We study a model for Traffic Shaping, Second Order Leaky Bucket, which consists of two leaky buckets to police the Sustained Cell Rate (SCR) and Burst Tolerance (BT) for each ATM connection. With this algorithm, ATM cells enter the first leaky bucket and the Cells conforming to the first leaky bucket enter the ATM network with Cell Loss Priority set to zero (CLP = 0). Any cell non-conforming to the first leaky bucket is sent to a second leaky bucket, and will be tagged CLP = 1 only if found to be non-conforming to the second leaky bucket. Cells conforming to the second leaky bucket are allowed to enter the ATM network with CLP = 0. We simulate the second-order leaky bucket for traffic shaping and show the effectiveness of second order leaky bucket in protecting the QoS experienced by connections passing through a common internodal link queue.

This thesis addresses a method to deduce the statistical bounds associated with the cell-transfer delay variations (CDVs) encountered by the cells of MPEG traffic, transmitted in the Asynchronous Transfer Mode (ATM) networks. This study focuses on: (1) Estimating CDV arising from multiplexing/switching for both constant bit rate (CBR) and variable bit rate (VBR) traffics via priority allocation based simulations. (2) Developing an information-theoretics based technique to get an insight of the combined BER-induced and multiplexing/switching-induced CDVs in ATM networks. Algorithms pertinent to CDV statistics are derived and the lower and upper bounds of the statistics are obtained via simulations in respect of CBR and VBR traffics. Ascertaining these bounds is useful in the cell admission control (CAC) strategies adopted in ATM transmissions. Inferential remarks indicating the effects of traffic parameters (such as bandwidth, burstiness etc.) on the values of the statistical bounds are presented, and scope for further work are presented.

This thesis addresses a method of deducing the statistical upper and lower bounds associated with the cell-transfer delay variations (CDVs) encountered by the cells transmitted in the asynchronous transfer mode (ATM) networks due to cell losses. This study focuses on: (1) Estimating CDV arising from multiplexing/switching for both constant bit rate and variable bit rate services via simulations. (2) Deducing an information-theoretics based new technique to get an insight of the combined BER-induced and multiplexing/switching-induced CDVs in ATM networks. Algorithms on the CDV statistics are derived and the lower and upper bounds of the statistics are obtained via simulations in respect of CBR and VBR traffics. These bounds bounds are useful in the cell-admission control (CAC) strategies adapted in ATM transmissions. Inferential remarks indicating the effects of traffic parameters (such as bandwidth, burstiness etc.) on the values of the statistical bounds are presented, and scope for further work is indicated.

Signal processing requires a steady flow of sampled data to be able to properly manipulate the signal to get the desired output. By using Asynchronous Transfer Mode (ATM) networks, it is possible to divide signal processing amongst a number of stations where each station can be specialized to a single function. Unfortunately, most commercially available ATM Network Interface Cards (NIC) only support message mode ATM Adaptation Layer 5 (AAL5) which is unsuitable to signal processing due to the delays of having to wait for an entire message to be formed prior to sending. It is shown that by using an ATM NIC using streaming mode AAL5, where cells are sent as soon as enough data to create an ATM cell of 48 bytes, leads to better quality signal processing. It is also shown that the message latency (time it takes for a message to traverse the network) is reduced by using streaming mode AAL5.

This thesis is concerned with an outline survey on the modern aspects of asynchronous transmission mode (ATM) telecommunication; also considered are certain specific issues such as the performance characteristics of ATM-based telecommunication networks under fluctuating workloads. Estimation of the stochasticity of the jitter associated with the cell-delay arising out of buffering within the network as well as due to signal-to-noise considerations which induce bit-errors and eventually lead to cell-losses manifesting as delay(s) in the cells being transmitted, is also addressed as a part of this thesis efforts. Relevant theoretical models are proposed, algorithms are developed and simulations are done on the above considerations. Based on these efforts, inferential conclusions are drawn and discussed. Scope for future research and developments towards engineering design of softwares suitable for practical applications by ATM networking planners and traffic forecasters is indicated. Merits and shortcomings of the models are proposed and algorithmic approaches pursued in the present study are also specified with relevant discussions.

In order to effectively transport digital compressed video over Broadband Integrated Services Digital Networks (B-ISDN) with Asynchronous Transfer Mode (ATM), the characteristics of the video source traffic should be understood. The nature of the video traffic depends primarily on the source, the content of the video and the coding algorithm that removes redundancies for efficient transmission over networks. In this study, video conference data encoded using a subband coding scheme, Digital Video Compression System (DVCS), is analyzed to determine its characteristics. Several video traffic sources are multiplexed through an ATM network node with limited capacity and the performance of this environment is evaluated by using a simulation technique. The simulation results are presented for the performance measures for varying traffic and network conditions.

Broadband-ISDN Network Architecture and Signaling concepts are described with particular emphasis on Asynchronous Transfer Mode (ATM) technology. A consolidated view of various aspects of B-ISDN/ATM Access and Network Signaling functions, architecture and protocols is presented. Additionally, a suggested evolutionary growth path for the B-ISDN signaling protocols is summarized. This is followed by a high-level comparison of two protocols under consideration for network signaling. The result of this evaluation indicates that the requirements of network signaling are best met by a protocol stack based on SS7 concepts. Finally, a set of future B-ISDN and Multimedia services is presented in context of the requirements they would impose on the signaling protocols. Enhancements to the access signaling protocol are proposed for the support of multiconnection and/or multiparty calls. These extensions/enhancements consist of a set of simplified messages, information elements and procedures based on message flows. Evolution and backward compatibility to existing protocols are taken into account while developing the extensions.