ATM call control by neural networks
Joaquim José Santos Esteves Neves, Luı́s B. Almeida, Mário Jorge Leitão · 1993
The resource allocation in the Broadband Integrated Services Dig-ital Network (B-ISDN) must guarantee the quality of service nego-tiated with new and existing calls, taking into account the Asyn-chronous Transfer Mode (ATM) statistical characteristics. A qual-ity of operation function, characterizing the overall network perfor-mance, is proposed, and based on this function, it is introduced a new strategy for the admission control and routing of the ATM call connections. As it is shown by simulation results, feedforword Neu-ral Networks trained with the backpropagation algorithm, can learn the traffic patterns in previous traffic situations, and can be used to predict the quality of operation changes caused by each new call. 1