Paper Info
Title
Testing the performance of a neural networks-based adaptive call admission controller with multimedia traffic
Abstract
We propose an Artificial Neural Network (ANN) based technique to be used for adaptive Call Admission Control (CAC) in Asynchronous Transfer Mode (ATM) networks. The mechanism does not depend on the traffic descriptors registered between the terminal and the network during connection setup time, thus inaccuracies in these descriptors will not affect the controller decisions. The controller monitors the current traffic state and adapts its decisions according to it. Also it takes into account the Quality of Service (QoS) of each traffic class separately as it bases the admission decision on the individual cell loss probability of each traffic class not on the average cell loss probability for mixed traffic. We modified the “Leaky Pattern Table” on-line training method to enable the NN to capture traffic variations more accurately. The proposed controller was tested with multimedia traffic as ATM is expected to be used for multimedia applications and we took into consideration the effect of the node buffer, otherwise the CAC mechanism will be too conservative. Reported results prove that this reactive controller is much more effective than other CAC mechanisms that depend on the traffic descriptors.
Year
DOI
Venue
1999
10.1007/978-3-540-48765-4_11
IEA/AIE
Keywords
Field
DocType
multimedia traffic,admission controller,neural networks-based adaptive call,neural network,asynchronous transfer mode,quality of service,artificial neural network
Asynchronous communication,Control theory,Adaptive system,Call Admission Control,Computer science,Computer network,Quality of service,Real-time computing,Asynchronous Transfer Mode,Artificial neural network,Multimedia,Traffic policing
Conference
Volume
ISSN
ISBN
1611
0302-9743
3-540-66076-3
Citations 
PageRank 
References 
0
0.34
8
Authors
3
Name
Order
Citations
PageRank
Yasser Dakroury100.68
Ahmed Abd El Al2929.03
Osman Badr300.34