Multi-path routing based on load-balance for cognitive packet networks

Quality of service （QoS） routing algorithms have been hardly discussed in the scientific community, most previous work on QoS routing concentrates on the performance of the single route. Cognitive packet network （CPN） has been designed for providing QoS routing. In this paper, to balance the loads among networks, we present a multi-path routing algorithm based on load-balance （MPRLB）, which is carried out in two steps. The algorithm with low computational complexity is firstly applied to establish multi path routing for each source and destination node pairs （SD-pair） nodes in the network. Then, we propose the hopfield neural network algorithm, which is applied to improve the efficiency of the flow deviation method for fast flow allocation among the links of the network based on load balance. Extensive simulation results demonstrate that the proposed scheme significantly improves the performance compared with the existing scheme that ignores load balancing.