Narrowband emitter identification based on Gaussian mixture model

Narrowband emitter identification can be regarded as a special application of data clustering for identifying unknown narrowband emitters from measured feature parameters. In this paper, narrowband emitter identification is divided into two different stages. In the first stage, a competitive stop expectation-maximization (CSEM) algorithm is developed, which is based on Shapiro-Wilk test and minimum description length variant (MDL2) criterion. The Shapiro-Wilk test is used to derive a decision whether to split a component into another two or not. In order to avoid over-splitting, the MDL2 criterion is employed as a competitive stop condition. The CSEM only employs the estimated elevation and azimuth angles at all the signal-occupied frequency bins as feature parameters. The frequency information implied in each cluster is not exploited sufficiently. So in the second stage, a postprocessing algorithm is introduced based on the implied frequency information. The experimental results show that the proposed CSEM algorithm has an increased capability to find the underlying model, while maintaining a low execution time. Combining CSEM and postprocessing algorithm, the narrowband emitter identification algorithm is able to determine the number of received narrowband emitters with high classification correctness and successfully estimate their characteristics. © 2010 Binary Information Press.