Optimal Methods for Classification of Digitally Modulated Signals

Abstract

This report summarizes the work done under the in-house effort titled "Optimal Methods for Modulation Classification." The effort revisited the problem of blind demodulation for develop classification algorithms for wider set of signals types. Two methodologies were used: Likelihood Ratio Test and inference methods using Expectation Maximization. A simple case of the likelihood was developed for spread spectrum. Also a minimum divergence approach was developed for classifying MPSK and QAM. Future work will refine these two proposed concepts. Finally, this research explored MIMO codes design. The purpose was to design codes that exceeded the performance of the Full-Rate, Full Diversity Golden Codes. It was possible to find linear codes with higher spectral rate, but also higher average power. This concepts will be refined as part of future work.