On-demand multipath protocol with adaptive routing method by back propagation neural networks

摘要

Two representative multipath routing protocol strategies have been utilised currently, one is described as primary routing protocol and the other is load-balancing protocol. In the former routing protocol, the source selects the pre-computed shortest or fastest route as the primary one for the following data transmission. On the other hand, the latter routing protocol utilises multiple routes in rotation to equalise transmission load in the network. However, these solutions suffer during high mobility since they are lacking in global perspective for the following data transmission, resulting in low packet delivery ratio and prolonged delay time. Hence, to find the optimum routing protocol strategy, we present an adaptive multipath routing protocol by means of applying golden section search and back propagation neural networks. We evaluated our protocol using Omnet simulator. Simulation results show that the proposed solution has a good real-time performance than those of other protocols.