Adaptive radial basis function diversity combiner for multipath channels

The Letter demonstrates the advantages of applying nonlinear signal processing to the problem of spatial diversity combining in digital communication channels suffering from severe intersymbol interference (ISI) due to multipath propagation. It is shown that a radial basis function (RBF) network can outperform an optimum linear combiner by approximating a Bayesian classification operation. Such a network can rapidly adapt to an unknown channel using a simple adaptive algorithm.