Using artificial neural networks to improve performance in a wireless ATM network

This paper examines some of the limitations of existing handover schemes and the potential performance and efficiency gains that can be achieved through hybrid variants that utilise neural networks. The virtual connection tree (VCT) is chosen as a candidate handover scheme because it statistically demonstrates a fast handover methodology that uses resource pre-allocation to maintain the QoS requirements for wireless ATM traffic. A hybrid multicast re-routing; scheme is introduced which aims to utilise the available bandwidth more efficiently and consequently to handle a higher number of concurrent calls.