Chaotic neural network with white noise for broadcast scheduling problems in packet radio networks

Transient chaotic neural network (TCNN) with chaotic simulated annealing (CSA), which can control the state searching in a fractal region, has a higher searching efficiency. However, the TCNN is not guaranteed to find the globally optimal solutions because CSA has completely deterministic dynamics. Contrary to CSA, stochastic simulated annealing (SSA) can search a globally optimal solution with probability one if the annealing speed is sufficiently slow. The white noise is one of the most common stochastic model and used widely in the engineering. In order to retain the excellent optimization property of SSA, the white noise is proposed into the TCNN. And the improved neural network is applied in the broadcast scheduling problems (BSP). The simulation results show that as long as we control the white noise in a proper range, we will be more likely to find an optimal or near-optimal time-division multiple-access (TDMA) frame structure with a minimal average delay time.