Immune System Assisted Radial Basis Function Network for OFDM System Channel Tracking in Dynamic Environments

In this paper, we develop an immune system assisted radial basis function (RBF) neural network for wireless channel tracking in pilot-added orthogonal frequency-division multiplexing (OFDM) systems. The RBF network is applied to approximate and follow the rapid change of channel response via online training of its output-layer weights. Meanwhile the network structure is adjusted in the assist of the immune system to stay optimal against environmental changes. The computational complexity is well under control, as the activation of immune system is much less often in comparison with the updating of network weights. The simulation result demonstrates that our algorithm is high-efficient, robust and immune to network overfitting especially in highly noisy environment.