A multilevel Hopfield neural network for OFDM system with phase noise

Phase noise causes common phase error (CPE) and intercarrier interference (ICI) in orthogonal frequency division multiplexing (OFDM) system. To mitigate the effects of phase noise, many schemes have been proposed, but they are either computationally complex or at the price of sacrificing spectral efficiency. In this paper, a continuous Hopfield neural network (HNN) is designed for OFDM signal receiver as a component to mitigate the effects of phase noise. This component does not require any special assistance from transmitter side, but it should be used in conjunction with phase noise matrix estimation. HNN, through self-evolution, performances likelihood test with possible symbols, and outputs a symbol for which the likelihood function has a relatively large value. The performance of the proposed HNN based detection is evaluated via computer simulations and compared to both the conventional detection and a single carrier system. It is shown that the HNN based detection can significantly improve the bit error rate (BER) performance in the presence of phase noise.