Complexity and Performance of Spectrally E cient Continuous Phase Modulation

Spectrally e cient Continuous Phase Modulation for a speci c narrow spectrum requirement mask and a prescribed information bit rate has been designed and analyzed in terms of the minimum Euclidean distance and simulated symbol error probability for the AWGN channel. The performance of a low complexity sub-optimum detector based on an approximate lter bank and a reduced state Viterbi decoder for these CPM schemes is evaluated. Furthermore, convolutionally encoded CPM has been investigated, and an optimization of the minimum Euclidean distance based on optimized phase pulses and modulation indices has been performed. The reduced complexity detector performance is close to the optimum detector, except when strong adjacent channel interference is present. Surprisingly, the performance of convolutionally encoded CPM is not better than uncoded CPM. However, improved performance is obtained by a phase pulse and modulation index optimization, and the sensitivity to adjacent channel interference for the sub-optimum detector is reduced.