CNN Approach to NP-complete Problems In Communication System

Cellular neural network is very suitable to solve communication problems because of its local connection. This paper summarizes a new approach for CNN's template design based on energy function, which is valid to such problems as equalizer and multi-user detector. This approach can combine the objective function of communication problems with CNN's energy function, and then obtain reasonable CNN's template. We also discuss the learning algorithm and hardware annealing joined with the approach. Simulation results prove that this method is valid to a class of communication problems