A bandit algorithm for multi-path routing in the emergency convergence network

In the convergence network which consists of cognitive self-organization network and mobile cellular network (MCN), the emergency service requires the network with higher reliability and quality of service (QoS). However, the time-variant factors such as spectrum, channel, network topology, node energy, as well as the gateway performance make the stochastic optimization problem of emergency service complex. In this paper, we create an optimum model which is based on the restless bandit theory[1], and propose an optimum algorithm for multi-path routing selection by considering the routing cross and overlap. In our algorithm, the multi-mode gateways are regarded as the key nodes to construct the routes between cognitive self-organization network and mobile cellular network. The cross and overlapped routes are taken into consideration to loosen and expand the routes set for selection. The communication process is divided into a few path decision stages, and some paths for multicast[2][3] communication are activated according to their real-time status in each stage. The trade-off between delay and energy efficiency is made for multi-path selection. Simulation results verify that our algorithm owns outstanding performance on decreasing delay and improving energy efficiency.