A binary neural network approach for link activation problems in multihop radio networks

This paper presents a binary neural network approach for link activation problems in multihop radio networks. The goal of the NP-compIete problems is to find a conflict-free link activation schedule with the minimum number of time slots for specified communication requirements. The neural network is composed of NX At binary neurons for scheduling jV links in M time slots. The energy functions and the motion equations are newly defined with heuristic methods. The simulation results through 14 instances with up to 419 links show that the neural network not only surpasses the best existing neural network in terms of the convergence rate and the computation time, but also can solve large scale instances within a constant number of iteration steps.