Using Hopfield network for multiuser signal detection

On principle that Hopfield neural networks can be used to solve combinatorial optimization problems, we propose the implementation of a multiuser signal detector with a Hopfield network and present a kind of Hopfield-based multiuser signal detector for synchronous DS/CDMA (direct sequence code division multiple access) systems. The essence of this scheme is establishing the relations between the likelihood function for optimum multiuser detection in synchronous DS/CDMA systems and an energy function for a Hopfield network. Simulation results show that the novel detector outperforms by far the conventional detector and has the performance comparable to that of the optimum detector at much lower computational cost.