Objective
Recently, owing to the great variety of required services, it became an urgent necessity in the field of data transmission technologies, to yield an integrated telecommunication network throughout Europe. One of the most prominent networks providing integrated services is the so-called Asynchronous Transfer Mode (ATM) network. In spite of its quickly growing importance, some aspects of this network are still far from standardisation. In that respect, one of the most important open questions of ATM networks is the call admission control (CAC) problem and its implementation on ATM switching nodes.
Since CAC basically determines the efficiency of resource management in ATM networks (i.e. how many users can simultaneously use the same medium for communication), the search for quick and optimal algorithms is of great practical importance. Besides applying an effective CAC algorithm, the network should police the users in terms of checking the deviation of their real traffic parameters from the declared ones. In case the users surpass their declared traffic parameters, a punitive action must be taken, which is performed by a network policing algorithm. The speeds of detection and punitive action of the policing algorithm contribute significantly to the overall quality performance of the network. On the other hand, to test the efficiency of newly introduced CAC or network policing algorithms a quick performance evaluation is needed.
As a consequence, the aim of the project is to develop new dynamic CAC and network policing methods, based on adaptive algorithms in order to ensure a flexible network management implemented by neural networks. To obtain an efficient CAC, a statistical study on the resource management problem will be carried out by applying the theory of statistical inequalities on the one hand, and the theory of neural networks on the other. The latter one will be of considerable help to provide an adaptive architecture based on its learning and approximation capabilities. This investigation also involves some practical research and simulations performed on ATM networks.
Obtaining an optimal and generally implementable CAC algorithm, which can possibly qualify for standardisation, will be to the benefit of all European countries. In this sense, the possible achievements of the project can be regarded as advancing towards a "European Information Highway", which supports a more or less unrestricted flow of information and associated services, regardless of the user's geometrical location. In particular, the results may serve to expand the Central and Eastern European teleworking facilities and to strengthen research links with the European Community, as well. By moulding the theoretical research (statistical resource management and neural network theory) into a practical application (ATM networks, CAC) the project is expected to provide theoretical as well as practical results bearing a great deal of relevance in the field.
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
CSC - Cost-sharing contractsCoordinator
3001 Leuven (Heverlee)
Belgium