MMSE cyclic equalization

In digital communications, conventional linear and decision feedback equalizers are optimal for stationary signals. However, modulated signals are polycyclostationary (PCS): their correlation functions are time-polyperiodic. Cyclic equalization structures optimized by the MMSE (minimum mean square error) criterion are proposed. The proposed structure is based on a multidimensional input. The latter is built from the received signal possibly shifted by its cyclic frequencies. We show by simulation that the proposed cyclic structures give better performance in terms of binary error rate than the conventional equalizers. The paper is organized as follows. First, we review some background topics on cyclostationarity and optimum filtering of PCS signals. Then we calculate the expressions of the forward and backward filters for the cyclic DFE and deduce the expression of the linear cyclic equalizer. Finally, simulation results of conventional and cyclic equalizers are given for BPSK and MSK signals jammed by other modulated signals. The new cyclic structures are particularly suitable for FDMA networks disturbed by adjacent channel interference (ACI). They can lead to a significant improvement in the number of communication channels within a given bandwidth. Besides, cyclic structures can also be used for transmission links disturbed by jammers that are modulated signals with the same or different baud rates.