Classification of digital modulation schemes using linear and nonlinear classifiers
Geisinger, Nathan P.
Fargues, Monique P.
Robertson, Ralph C.
MetadataShow full item record
The potential benefits of automated detection of digital modulation types have made it a continuing topic of research for many years. Commercial systems could be made more interoperable and military sensors could send demodulated products for analysis, to name just two. Noisy channels and multipath fading environments continue to make this a challenging problem. This thesis applies classification algorithms that have been used in other applications. Nine different digital modulation schemes are considered. The criteria for selecting higher-ordered moments and cumulants as features for discrimination are discussed. An overview of the classification algorithms considered is provided, as well as the statistical models for noisy channels. Results show that the scheme proposed here works well in AWGN channels and in moderate fading conditions.
Approved for public release; distribution is unlimited