Dilip Sarkar

Optimal broadcasting on the star graph

The star graph has been show to be an attractive alternative to the widely used n-cube. Like the n-cube, the star graph possesses rich structure and symmetry as well as fault tolerant capabilities, but has a smaller diameter and degree. However, very few algorithms exists to show its potential as a multiprocessor interconnection network. Many fast and efficient parallel algorithms require broadcasting as a basic step. An optimal algorithm for one-to-all broadcasting in the star graph is proposed. The algorithm can broadcast a message to N processors in O(log/sub 2/ N) time. The algorithm exploits the rich structure of the star graph and works by recursively partitioning the original star graph into smaller star graphs. In addition, an optimal all-to-all broadcasting algorithm is developed. >

Modeling full-length video using Markov-modulated gamma-based framework

All traffic models for MPEG-like encoded variable bit rate (VBR) video can be broadly categorized into 1) data-rate models (DRMs) and 2) frame-size models (FSMs). Almost all proposed VBR traffic models are DRMs. DRMs generate only data arrival rate, and are good for estimating average packet-loss and ATM buffer overflowing probabilities, but fail to identify such details as percentage of frames affected. FSMs generate sizes of individual MPEG frames, and are good for studying frame loss rate in addition to data loss rate. Among three previously proposed FSMs: 1) one generates frame sizes for full-length movies without preserving group-of-pictures (GOP) periodicity; 2) one generates VBR video traffic for news videos from scene content description provided to it; and 3) one generates frame sizes for full-length movies without preserving size-based video-segment transitions. In this paper, we propose two FSMs that generate frame sizes for full-length VBR videos preserving both GOP periodicity and size-based video-segment transitions.First, two-pass algorithms for analysis of full-length VBR videos are presented. After two-pass analysis, these algorithms identify size-based classes of video shots into which the GOPs are partitioned. Frames in each class produce three data sets, one each for I-, B-, and P-type frames. Each of these data sets is modeled with an axis-shifted Gamma distribution. Markov renewal processes model (size-based) video segment transitions. We have used QQ plots to show visual similarity of model-generated VBR video data sets with original data set. Leaky-bucket simulation study has been used to show similarity of data and frame loss rates between model-generated VBR videos and original video. Our study of frame-based VBR video revealed that even a low data-loss rate could affect a large fraction of I frames, causing a significant degradation of the quality of transmitted video.

Methods to speed up error back-propagation learning algorithm

Error back propagation (EBP) is now the most used training algorithm for feedforward artificial neural networks (FFANNs). However, it is generally believed that it is very slow if it does converge, especially if the network size is not too large compared to the problem at hand. The main problem with the EBP algorithm is that it has a constant learning rate coefficient, and different regions of the error surface may have different characteristic gradients that may require a dynamic change of learning rate coefficient based on the nature of the surface. Also, the characteristic of the error surface may be unique in every dimension, which may require one learning rate coefficient for each weight. To overcome these problems several modifications have been suggested. This survey is an attempt to present them together and to compare them. The first modification was momentum strategy where a fraction of the last weight correction is added to the currently suggested weight correction. It has both an accelerating and a decelerating effect where they are necessary. However, this method can give only a relatively small dynamic range for the learning rate coefficient. To increase the dynamic range of the learning rate coefficient, such methods as the bold driver and SAB (self-adaptive back propagation) were proposed. A modification to the SAB that eliminates the requirement of selection of a good learning rate coefficient by the user gave the SuperSAB. A slight modification to the momentum strategy produced a new method that controls the oscillation of weights to speed up learning. Modification to the EBP algorithm in which the gradients are rescaled at every layer helped to improve the performance. Use of expected output of a neuron instead of actual output for correcting weights improved performance of the momentum strategy. The conjugate gradient method and self-determination of adaptive learning rate require no learning rate coefficient from the user. Use of energy functions other than the sum of the squared error has shown improved convergence rate. An effective learning rate coefficient selection needs to consider the size of the training set. All these methods to improve the performance of the EBP algorithm are presented here.

Transport Layer Protocols for Cognitive Networks

The cognitive radio networks or CogNets poses several new challenges to the transport layer protocols, because of many unique features of cognitive radio based devices used to build them. CogNets not only have inherited all features of wireless networks, but also their link connections are intermittent and discontinuous. Exiting transport layer protocols are too slow to respond quickly for utilizing available link capacity. Furthermore, existing self-timed transport layer protocols are neither designed for nor able to provide efficient reliable end-to-end transport service in CogNets, where wide round trip delay variations naturally occur. We identify (i) requirements of protocols for the transport layer of CogNets, (ii) propose a generic architecture for implementing a family of protocols that fulfill desired requirements, (iii) design, implement, and evaluate a family of best-effort transport protocols for serving delay-tolerant applications. Results obtained from ns-2 network simulator show that the proposed protocols have potential for significantly improving end-to-end throughput. For instance, at 1% and 5% packet loss rates one of the proposed protocol has shown about 21% and 95% increase in end-to-end throughput for file transfer application.

Convergence in the calculation of the handoff arrival rate: a log-time iterative algorithm

Modeling to study the performance of wireless networks in recent years has produced sets of nonlinear equations with interrelated parameters. Because these nonlinear equations have no closed-form solution, the numerical values of the parameters are calculated by iterative algorithms. In a Markov chain model of a wireless cellular network, one commonly used expression for calculating the handoff arrival rate can lead to a sequence of oscillating iterative values that fail to converge. We present an algorithm that generates a monotonic sequence, and we prove that the monotonic sequence always converges. Lastly, we give a further algorithm that converges logarithmically, thereby permitting the handoff arrival rate to be calculated very quickly to any desired degree of accuracy.

A Concurrent Multipath TCP and Its Markov Model

Concurrent Multipath Transport (CMT) has been of increasing interest recently for aggregating bandwidth, balancing load, and increasing reliability. In this paper we propose a concurrent multipath Transport Control Protocol (cmpTCP), which is an extension of TCP New Reno and Steam Control Transmission Protocol (SCTP). The proposed cmpTCP has been implemented by altering and extending library-based open-source for SCTP. The proposed cmpTCP manages transport control parameters of all paths simultaneously; a scheduler concurrently dispatches packets over all the paths from a common transmission queue. However, one virtual retransmission queue is maintained for each path, and the receiver sends acknowledgment for a packet on the same path on which the packet is received. A Markov model for estimation of expected window size of each path has been proposed. Next, the proposed model has been utilized to derive an expression for computing average data transmission rate when the proposed cmpTCP is used. To validate performance of the proposed model, two host computers were connected through two independent network emulators. Each network emulator runs on a separate computer, and the network parameters are adjusted independently. We used our cmpTCP for transferring files from one host to the other. Comparison of experimental results with model predicted results shows excellent agreement.

Randomness in generalization ability: a source to improve it?

Among several models of neurons and their interconnections, feedforward artificial neural networks (FFANNs) are most popular, because of their simplicity and effectiveness. Difficulties such as long learning time and local minima may not affect FFANNs as much as the question of generalization ability, because a network needs only one training, and then it may be used for a long time. This paper reports our observations about randomness in generalization ability of FFANNs. A novel method for measuring generalization ability is defined. This method can be used to identify degree of randomness in generalization ability of learning systems. If an FFANN architecture shows randomness in generalization ability for a given problem, multiple networks can be used to improve it. We have developed a model, called voting model, for predicting generalization ability of multiple networks. It has been shown that if correct classification probability of a single network is greater than half, then as the number of networks in a voting network is increased so does its generalization ability. Further analysis has shown that VC-dimension of the voting network model may increase monotonically as the number of networks in the voting networks is increased.

A Call-Admission Control (CAC) Algorithm for Providing Guaranteed QoS in Cellular Networks

Future broadband wireless access systems are expected to integrate various classes of mobile terminals (MTs), each class with a different type of quality of service (QoS) requirement. When the load on a wireless network is high, the guarantee of QoS for each class of MTs is a challenging task. This study considers two classes of MTs—profiled MTs and nonprofiled or regular MTs. It is assumed that profiled users require a guaranteed QoS. The measure of QoS is the probability of forced termination of a call that was allowed to access the network. Two previous handoff prioritization schemes—(i) prerequest scheme and (ii) guard channel scheme—decrease handoff failure (and hence forced termination). In this work, we compare and contrast both the schemes through extensive simulation and we find that neither guard channel nor channel prerequest scheme can guarantee a desired level of QoS for the profiled MTs. We then propose a novel call-admission control (CAC) algorithm that can maintain any desired level of QoS, while the successful call completion rate is very high. In the proposed algorithm, the new call arrival rate is estimated continuously, and when the estimated arrival rate is higher than a predetermined level, some new calls are blocked irrespective of the availability of channels. The objective of this new call preblocking is to maintain a cells observed new call arrival rate at no more than the predetermined rate. We show that the proposed method can guarantee any desired level of QoS for profiled users.

Cost and time-cost effectiveness of multiprocessing

Speedup and efficiency, two measures for performance of pipelined computers, are now used to evaluate performance of parallel algorithms for multiprocessor systems. However, these measures consider only the computation time and number of processors used and do not include the number of the communication links in the system. The author defines two new measures, cost effectiveness and time-cost effectiveness, for evaluating performance of a parallel algorithm for a multiprocessor system. From these two measures two characterization factors for multiprocessor systems are defined and used to analyze some well-known multiprocessor systems. It is found that for a given penalty function, every multiprocessor architecture has an optimal number of processors that produces maximum profit. If too many processors are used, the higher cost of the system reduces the profit obtained from the faster solution. On the other hand, if too few processors are used, the penalty paid for taking a longer time to obtain the solution reduces the profit. >

Hypercube connected rings: a scalable and fault-tolerant logical topology for optical networks

A new, fault-tolerant, scalable, and modular virtual topology for lightwave networks employing wavelength division multiplexing is proposed. The proposed architecture is based on a hypercube connected ring structure that enjoys the rich topological properties of a hypercube, but it also overcomes one of its drawbacks. In a hypercube, the nodal degree increases with the number of nodes. Hence, the per-node cost of the network increases as the network size grows. However, in a hypercube connected ring network (HCRNet) the nodal degree is small and it remains constant, independent of the network population. A HCRNet, like a hypercube, is perfectly symmetric in the sense that the average internodal distance in an N-node HCRNet is the same from any source node. Its average internodal distance is in the order of logN and it is comparable to other regular structures such as the Trous and ShuffleNet. The HCRNet is based on the Cube Connected Cycle (CCC) interconnection pattern proposed for multiprocessor architectures. However, the HCRNet improves on CCC by rearranging its hypercube links, which results in a significantly lower average internodal distance. In this paper we present the structural properties of HCRNet, and address the issues of scalability, and fast routing in complete as well as incomplete HCRNet.