Global rational approximation for call admission control in ATM networks

This paper presents a framework for call admission control (CAC) in ATM networks based on case-based reasoning (CBR). CBR is used to correct the error estimation of the required bandwidth computed by conventional call admission control schemes, which were shown to overestimate the required bandwidth. This leads to bandwidth wastage and increased call rejection. A CBR-based system is proposed to characterize the traffic that may affect the cell loss ratio (CLR) of the network. The proposed system consists of two phases, an off-line phase and an on-line phase. In the off-line phase, the system constructs an initial explanation for having a high cell loss rate (failure cases) resulting from accepting a larger number of calls than desired. In the on-line phase, the system uses its explanations to make a decision of accepting or rejecting a new call. If a failure explanation is applicable for the new call, then the new call is rejected. Otherwise, the new call is accepted. The learning arises from receiving a feedback of the resulting CLR to evaluate the decision made by the proposed system and to update the explanation previously made. The performance of the scheme was shown to be superior compared to conventional schemes in terms of system utilization and call blocking ratios.

In this paper the utilisation of the Probability of Congestion (PC) as a bandwidth allocation decision parameter is presented. We assume short buffers at the switch nodes to cope with cell level multiplexation contention (“bufferless” environments). Therefore, delay and cell delay variations are strongly bounded. Moreover, the Cell Loss Ratio (CLR) becomes the critical performance parameter. The validity of PC utilisation is compared with quality of service parameters in bufferless environments. The convolution algorithm is an accurate approach for Connection Admission Control (CAC) in ATM networks with small buffers. However, the convolution approach has a considerable computation cost, in terms of calculation and memory. To overcome these drawbacks, a new method of evaluation is proposed and analysed: the Enhanced Convolution Approach (ECA). In complex scenarios, with ECA, PC calculation can be carried out in real time while maintaining the desired accuracy. Several experiments have been carried out to compare the demanded bandwidth evaluated by: analytical methods, simulations and measurements in actual ATM switches. The main contribution of this paper is the proposal and analysis of ECA to the PC-evaluation for use in CAC schemes

An adaptive scheme for admission control in ATM networks

This paper is concerned with dynamic resource allocation in ATM networks. One of the essential issues in designing and configuring an ATM network is guaranteeing the quality of service for different connections. This needs more advanced functions to monitor the current situation, determine the required resource, allocate the resource and modify it. In this paper we introduce a new dynamic resource allocation metric based on probabilistic tools which serves as a monitoring mechanism of resource allocation in ATM systems. © 1998 John Wiley & Sons, Ltd.

We propose and analyze a new call admission controller for ATM networks using neural networks (NN). The proposed model is based upon real time measurements of the traffic via a simple parameter, which is the number of cells arriving during the measurement interval. The length of the measurement interval and the number of traffic samples within, are selected to capture the variability properties of the traffic. A neural network controller is then trained to learn the long term correlation properties of the traffic which is essential for effective statistical multiplexing and bandwidth allocation. A large set of training data representing multiservice traffic patterns with multiple QOS requirements is used to ensure that the controller can generalize and produce accurate results when confronted with new test data. The reported results prove that the neural network approach is effective in estimating the bandwidth requirements, when compared to other traditional methods that are based upon an algorithmic approach. This is, primarily, due to the unique learning and adaptive capabilities of neural networks enable them to approximate any non-linear function from previous experience. Evidently, such unique capabilities are the reasons for proposing the use of neural networks to solve many of the problems encountered in the design of ATM networks.

A new approach to determining the admissibility of a new call request in a buffered ATM network is developed. In this approach, all traffic presented to the network is assumed to be statistically multiplexed sources. This leads to statistical interactions among different traffic sources. The extent of multiplexing gain (spare capacity) is efficiently estimated and the bandwidth (resource) allocation is traded off in an optimal call admission control policy. The algorithms presented in this paper are tested by simulation and their effectiveness is investigated.

We consider large loss networks with fixed routing and multi-rate traffic. We use the Laplace method and standard results of multidimensional Gaussian distribution to obtain the upper bounds for the blocking probabilities of new calls under uniform light loading conditions. This is the loading regime of interest for many practical applications such as admission control in ATM networks. The main advantage of our approach is that the complexity does not scale with the capacity of the system, so computation are extremely rapid in any case. We verify our method on a small network example. Comparison with simulation results show that we get good upper bounds.

Recent research has shown that fluid-flow queueing models accurately determine the burst level component of the ATM multiplexer process, and fluid models therefore constitute an efficient tool for the investigation of admission control in ATM networks. This article first gives a review of Markov-driven fluid queues with special attention given to the heterogeneous on/off model. Then an approximation procedure in which the original heterogeneous arrival process is replaced by a homogeneous birth-death arrival process is presented. The substitution is based on matching statistical parameters. Finally, the approach is evaluated by numerical examples. >

In this paper, we consider large loss networks with fixed routing and multi?rate traffic. We use single?link formulae and standard results on multidimensional Gaussian distributions to obtain upper bounds for blocking probabilities of new calls under light up to critical loading conditions. This is the loading regime of interest for many practical applications such as admission control in ATM networks. The main advantage of our approach is that the complexity does not scale with the size of the system, making it numerically attractive. Comparison with simulation results show that we get good upper bounds. We conclude by discussing the correlation between links in a network.

Cell Loss Ratio estimation is a crucial technology in call admission control and traffic engineering in ATM networks. Cell arrival process and analytical approach had been investigated in recent years. However, an effective and practicable CLR estimator is still a challenge. Based on the analysis of cell loss problem, an improved simple traffic model and a new algorithm are presented in this paper, to estimate CLR of ATM network multiplexed with heterogenous traffic classes services rapidly. Traffic model is constructed with standard parameters, so it is easy to use in practical situations. The new algorithm runs quickly enough to respond the call real-time. Simulation results show that accuracy, complexity and robustness of algorithm are ideal to be utilized in real network.

Estimating cell loss probability (CLP) in real time for achieving satisfactory call admission control (CAC) remains an important research problem. In estimating CLP which is often specified to be less than 10/sup -9/, fuzzy based techniques have been found to be more promising than crisp statistical techniques. In this paper, we propose a novel fuzzy logic algorithm based on the /spl alpha/-cuts of equivalence relations to estimate CLP in real time for self-similar ATM networks. The /spl alpha/-cuts of fuzzy sets are used to construct the membership functions of the input and output variables and to induce necessary fuzzy rules. The proposed fuzzy approach is validated by comparing the estimated values of CLP with the theoretical values. The new approach not only estimates accurate real time CLP, but also achieves it by using fewer theoretical data.

In this paper a global control system for admission control, routing and bandwidth allocation in the ATM environment is presented. The traffic is divided into classes, characterized by statistical parameters and by quality of service (QoS) requirements, as the rates of lost and delayed cells. At connection set-up, a strategy decides if a connection request has to be accepted or rejected and which is the best route that all the cells of that connection have to follow. The decision about acceptance guarantees the QoS requirements for each class, while the routing strategy is aimed at minimizing the number of blocked calls in the network. The whole system provides also a mechanism to achieve a fair and adaptive bandwidth allocation for each traffic class; the bandwidth allocation control has a longer intervention period than the admission and routing control. In fact, the global strategy can be regarded as a hierarchical one. Several simulation results are reported, to verify the effectiveness both of the admission and routing mechanism and of the allocation strategy.

Integrated dynamic distributed routing and admission control in ATM networks

Summary form only given. The problem of flow control in ATM networks is a difficult one due to the uncertainties about the flow rates of the data streams coming from various sources of traffic at different nodes of the network. The flow control problem consists of multiplexing the ATM traffic and it is a challenging statistical management strategy where different patterns are directed into the communication channels. During the multiplexing stage, some prescribed performance constraints on the communication parameters should be strictly satisfied. The real motivation for considering an unconventional control technique for flow control in the ATM networks is because with the growing demand for the traffic of the integrated services (including voice and data communications), reliable decisions should be taken in real time. We consider the problem of estimating the statistical parameters of an intelligent control scheme for smoothing the ATM networks based on a neural network architecture. The neural network considered for carrying out the task of flow control runs the statistical estimation as an iterated map. The reason for this architectural design is the need to have a robust estimation scheme, where the convergence to an attractor of the neural dynamics can be enforced to hold in a small number of iterations of the statistical mapping. The neural network is symmetrically connected and this choice reflects a kind of tradeoff between the speed of convergence and the stability of the states during the successive states of the dynamical relaxation. A numerical experiment describing the dynamic performances of the neural controller for the ATM flow control is presented.

The statistics of cell loss at both burst and cell levels are studied by simulation based on the configuration of the Singapore One ATM network. A virtual connection through the Singapore One network is modelled as a tandem queueing network to which mixtures of homogeneous discrete-time burst–silence traffic with different burst lengths are offered. Attention is focused on a designated traffic stream as it goes through the tandem queueing network. In the simulation, the authors have not introduced the ‘independence assumption’ between the adjacent queues, i.e. the actual tandem queueing network is used to obtain the performance measures for the designated traffic stream. Cell loss probability and high-order cell loss distributions such as cell loss-free run distribution and cell loss pattern distribution are obtained. These results provide valuable information for the estimation of end-to-end overall quality of service (QoS), design of the cell policing mechanism and flow control schemes in the ATM networks.