This paper addresses the problem of modelling and analysing an interior node in IETF's DiffServ services model. Specifically it is concerned with the performance of the DiffServ assured forwarding service category in presence of a premium service class. In this model, the network node shares its outgoing link capacity between a premium service representing the expedited forwarding (EF) per-hop behavior, and two classes of assured service, that represent two classes of the assured forwarding (AF) per-hop behavior. It this paper, the traffic is modelled as Markov modulated fluid sources, and we focus on a system where out of profile traffic is dropped at the edge of the network thus both AF queues support only one drop precedence. Using a decomposition approach, approximations, and spectral analysis, we are able to derive upper and lower bounds on the tail of the distribution of the buffer content for both AF classes given a generalized processor sharing scheduling is used to differentiate the two classes. Such approximate analysis of the interaction between traffic classes can help to achieve a better understanding of this type of networks; enables the provision of throughput differentiation as defined by the AF PHB through the GPS scheduler while quantifying delay; and finally helps simplify greatly the design of bandwidth brokers that do not rely on long term bandwidth (over) provisioning.

In this paper, we propose a new call admission control scheme called dual threshold bandwidth reservation, or DTBR scheme. The main novelty is that it builds upon a complete sharing approach, in which the channels in each cell are shared among the different traffic types and multiple thresholds are used to meet the specific quality-of-service (QoS) requirements. We present a detailed comparative study based on mathematical and simulation models, and quantitatively demonstrate that the DTBR is capable of providing the QoS guarantee for each type of traffic, while at the same time leading to much better channel efficiency. We further show that the DTBR scheme with elastic data service can offer both service guarantee and service differentiation for voice and data services, and enhance the bandwidth utilization.

Layer transmission generates multiple layers for a video program, enabling a receiver to selectively subscribe to the layers commensurate with its bandwidth. It is an effective solution to the problem of bandwidth heterogeneity in video broadcasting. However, two important issues remain to be addressed. First, how does a receiver select the subset of layers to achieve the highest bandwidth utilization? Second, how does the sender optimally allocate the layer bandwidth to match the diverse bandwidth requirements from the receivers? We formally investigate these problems in noncumulative layered broadcasting, where any subset of the layers can be used to reconstruct the video. We formulate both the optimal layer subscription problem for a receiver and the optimal layer bandwidth allocation problem for the sender. We show that the former has an effective solution, while the latter is computationally intractable. Three efficient heuristic algorithms are then proposed for the allocation problem, and simulation results show that all of them significantly outperform nonadaptive allocation algorithms.

One of the key issues in the next generation wireless cellular networks is to support multiple services, in which call admission control policy has to guarantee the different quality of service (QoS) requirement from diverse applications, and at the same time ensure the scarce bandwidth be utilized efficiently. In this paper, we first build upon our proposal dual threshold bandwidth reservation (DTBR) scheme to integrate more realistic, variable data traffic, and them proceed to analyze the DTBR scheme using two-dimensional Markov process. Both analysis and simulation results indicate that, by keeping target handoff voice call dropping probability, adoption of elastic data traffic model can reduce handoff voice call dropping probability and improve overall system award; by adjusting the threshold values in the DTBR, we demonstrate that it is capable of satisfying different QoS requirements for voice and data traffic while achieving the maximum utility.

One of the key challenges in the design of bandwidth allocation policies for a multi-services mobile cellular network is to guarantee the potentially different quality of service (QoS) requirement from diverse applications, while at the same to ensure that the scarce bandwidth be utilized efficiently. Complete sharing (CS) and dynamic partition (DP) schemes have been shown as viable techniques for managing the bandwidth. However, there has been no study that compares their respective performance, which is the focus of this paper. Specifically, in this paper, through both analysis and simulation, we demonstrate that both schemes can achieve comparable performance by proper manipulation of control parameters. The tradeoff is that DP scheme can more easily achieve the target QoS requirement, at the expense of over-provisioning, thus can potentially lead to less channel efficiency when comparing to a CS based scheme.

In layered video broadcasting, if adaptation is performed only by receivers, significant mismatches between a receiver's expected bandwidth and the actually delivered bandwidth could occur as the adaptation unit is a coarse-grained layer. The paper shows that fine-grained sender adaptation, as a complement to receiver adaptation, can significantly decrease these mismatches. A formal study on optimal session and layer bandwidth allocations for sender adaptation in a broadband cellular network is carried out. The most fundamental issues associated with layered video broadcasting, including system utility and the overhead of layering, are considered. A polynomial-time algorithm is derived for optimal allocation. Experimental results show that it can significantly improve the system utility compared to static allocation algorithms.

We develop adaptive algorithms for multichannel (single-input-multiple-output, or SIMO) blind identification with both statistic and deterministic models. In these algorithms, the estimates are continuously improved while receiving new signals. Therefore, the algorithms can track the channel continuously and thus are amenable to real applications such as wireless communications. At each step, only a small amount of computation is involved. The algorithms are based on stochastic-approximation methods. The convergence properties of these algorithms are proved. Simulation examples are presented to show the performance of the algorithms.

The basic concepts of three branches of game theory, leader-follower, cooperative, and two-person nonzero sum games, are reviewed and applied to the study of the Internet pricing issue. In particular, we emphasize that the cooperative game (also called the bargaining problem) provides an overall picture for the issue. With a simple model for Internet quality of service (QoS), we demonstrate that the leader-follower game may lead to a solution that is not Pareto optimal and in some cases may be "unfair," and that the cooperative game may provide a better solution for both the Internet service provider (ISP) and the user. The practical implication of the results is that government regulation or arbitration may be helpful. The QoS model is also applied to study the competition between two ISPs, and we find a Nash equilibrium point from which the two ISPs would not move out without cooperation. The proposed approaches can be applied to other Internet pricing problems such as the Paris Metro pricing scheme.

Packet reservation multiple access (PRMA) is a media access control (MAC) protocol for wireless systems which can also be used as a packet control layer on top of TDMA. This paper analyzes the performance of PRMA for data traffic in a wireless communication system. A modification has also been proposed to improve the data transmission fairness under congested traffic condition. Markov model and computer simulations have been used to study the channel utilization, packet access time etc. of the data system. The analysis shows that the performance of the PRMA protocol for data traffic is satisfactory. The derived mathematical model provides an accurate tool for evaluating the system performance.

We study the blind separation problem of non-Gaussian source signals when there is only one desired signal to be recovered from its instantaneous mixtures with another source signals. Two sets of equations are built, both of which can be solved to blindly extract one non-Gaussian source signal. When there are n source signals, the number of equations in each equation sets is 3(n-1) that is much less than that for the separation of all source signals. The contribution of this paper lies in that it develops a principle for the design of blind single source separation algorithm.

We propose a direct blind zeroforcing approach to cancel intersymbol interference (ISI) in multiple user finite impulse response (FIR) channels. By selectively anchoring columns of the channel convolution matrix, we present two column-anchored zeroforcing equalizers (CAZE), one without output delay and one with a chosen delay. Unlike many known blind identification algorithms, these equalizers do not need an accurate estimate of the channel orders. Exploiting second-order statistics (SOS) of the received signals, they can retain preselected d columns in the channel convolution matrix (d is the number of users) and force the remaining columns to zero. CAZE can effectively equalize single-input-multiple-output (SIMO) systems and can reduce dynamic multiple-input-multiple-output (MIMO) systems into a memoryless signal mixing system for source separation. Simulation results show that the CAZE is not only effective for blind equalization of linear quadrature amplitude modulation (QAM) systems, but it is also applicable to the nonlinear GMSK modulation in the popular wireless GSM systems when computational cost severely limits the use of nonlinear methods such as the Viterbi algorithm.

The concept of voice and telephony over ATM (VTOA) has been widely studied. Although ATM benefits voice communications in many aspects, its packet-switched nature impairs the speech quality. We evaluate the speech quality with a focus on the impact of the cell delay variation (CDV). We build an experimental system to do the evaluations. We also define a new parameter, named cell delay variation accumulation length (CDVAL), to measure the impact of CDV on the speech quality. We observe that the speech quality mainly depends on the value of the CDVAL, but does not strongly depend on the other items, including the distribution of cell end-to-end delay, the position and even the number of the intermittence caused by the CDV. We find that there exists an acceptable threshold of the CDVAL. According to our experiments, to keep the speech in "good" quality (MOS/spl cong/3.5), the CDVAL value must be less than 3 ms without any speech cell loss; such a threshold drops to 2 ms if 5% of the speech cells are lost. The results indicate that a buffer is necessary for CDV compensation at the far end devices in VTOA. The results are also very helpful for the proper design of the network architecture and control algorithms, and is especially important for the design of the CDV compensation schemes, when ATM networks are used to provide voice and telephony services.

We introduce an analytical model to obtain the performance bounds for a single hop optical network based on WDM. We model the network under saturation traffic loading while ignoring the specific channel access scheme as a multi-class closed queueing network (BCMP network), for which a product-form solution for the steady-state probabilities exists. Therefore the exact solutions can be derived. In addition, we study the effect on the system performance of the number of stations in the network, the number of transmitters/receivers at each station, and the traffic distribution pattern.

Performance potential theory has proved to be a promising tool in optimizing the infinite-horizon Markov decision problem (MDP). So far, the research in this area is implicitly focused on a simple system with a single controller. In this paper, we consider the distributed controlled Markov chain, where the system consists of several individual control units and it evolves under the combined control of these nodes. Motivated by practical background, we investigate a structure of MDP with event-dependent decisions. We explore a notion of expanded Markov chain to map this problem to a traditional MDP model. In particular, we address ourselves to the complexity-reduction techniques to deal with the enlarged state space. For the distributed system where a particular node can only access partial system information, we develop some algorithms for decentralized potential estimation and policy iteration.

The performance evaluation of ATM switches is of paramount importance in the design and analysis of ATM networks. In this paper, we focus on the evaluation of the cell loss rate (CLR) and cell delay probability (CDP) in nonblocking ATM switches using computer simulations. In particular, we investigate the potential of using importance sampling techniques as an "superfast" alternative to conventional Monte Carlo simulation in finding the CLR and CDP in nonblocking ATM switches. We propose a "split switch" method to decouple the input and output queue behaviors, along with the notion of regenerative cycles. Numerical results will demonstrate that considerable computation cost can be saved using the proposed importance sampling techniques while maintaining a high degree of accuracy.

We propose a discrete-time queueing network model for space-division packet switches, in which the destinations of consecutive packets may be correlated, and different input traffic may have different destination distributions. We show that for switches under saturation, the steady-state probabilities have an approximate product-form solution. The solution is very accurate for real applications where the average packet sizes are often more than 10 times the number of bytes that can be transmitted in one time slot.>

The maximum throughput of an ATM switch is investigated in the presence of an offered load of multicell packets. For the case of input queueing coupled with a round-robin policy for transferring cells from inputs to outputs, the system is approximated by a product form queueing network. Under the assumption that packet lengths are described by random variables with discrete Coxian distributions, it is shown that the balance equations describing the behavior of the ATM switch approach those for a product form queueing network and that the steady-state probabilities of such an ATM switch approach the product-form solution as the cell length tends to zero. Last, a numerical investigation shows that the approximation yields good results, even when the packet lengths are not well described by Coxian distributions.>

The results of perturbation analysis pertaining to closed queuing networks are reviewed, showing that the basic principles of perturbation analysis can be explained using a dynamic point of view. It is shown that perturbation generation and perturbation propagation rules can be viewed as counterparts of the linearization theory of nonlinear continuous variable dynamic systems; the concept of perturbation realization describes the steady-state effect of a perturbation and reflects the special dynamic feature of a closed queuing network. The realization probability can be used to calculate the sensitivity of steady-state throughputs and some other sensitivities.>

Three kinds of estimates of the performance sensitivity of a stochastic system are discussed. The convergence properties of these estimates are investigated. The first estimate, using the time average of the derivative of the performance function calculated along a sample trajectory, is generally preferable when certain conditions hold for the performance function. The variance of the second estimate, using the same input random process, is much less than that of the third estimate, which uses tow different input processes. An example of a one-dimensional linear system with a quadratic performance function is given; it illustrates the general approach to verifying the conditions related to the first estimate for linear systems.>

A new approach for estimating the throughput sensitivity with respect to routing probabilities in queueing networks and flexible manufacturing systems is proposed. The approach converts the problem to a problem of calculating the sensitivity of throughput with respect to service distributions using perturbation analysis.

This paper is based on a recent technique called perturbation analysis of discrete event systems. The aims of the paper are two-fold. The first, is to bring this technique, which is still relatively new, to the attention of a wider audience, since the method is particularly relevant to practical systems optimization. The second, is to develop two new methods, the phantom customer method and the marked customer method, as extensions of the perturbation analysis approach. These methods enable efficient and accurate optimization of closed queueing networks with blocking and general service-time distributions. Useful practical applications of these methods would be for optimization of performance of computer systems, communication networks, and automated manufacturing complexes. The accuracy of the methods is illustrated by several numerical examples.