Average Likelihood Methods of Classification of Code Division Multiple Access (CDMA)

This final report summarizes the most significant findings under the in-house effort titled Average Likelihood For CDMA.The effort included refinements to the proposed average likelihood method discussed in AFRL-RI-RS-TR-2014-09-093.The refinements include a Taylors approximation of the log-likelihood function the development of algorithms to extract features, the inclusion of unbalanced CDMA cases and the extension of the average likelihood to complex cases of CDMA. The performance of the classifier is validated by the use of accuracy versus SNR curves. Classification of CDMA under a hypothesis expressed in terms of code length and active number of user is feasible by performing a simple preprocessing stage at the beginning of the classifier.