Detection of model order variations in frequency hopped systems

Model order information is important in parametric methods in array signal processing. Many model order estimation techniques assume a fixed model order within the observation duration. In frequency hopped (FH) systems, however, signals may appear to hop in and out of the observation frequency band if there is a bandwidth mismatch between the receiver and transmitters, which results in model order variations. In this paper, we propose a low complexity method to detect model order changes. Our detection method is based on the "soft-information"-the trace of signal correlation matrices obtained using a sliding-window approach. Simulations show that at low signal-to-noise ratio (SNR), the proposed method offers better performance than detection based on "hard-information" obtained by commonly used source enumeration techniques such as SVD and MDL. The method is developed in the context of FH systems, but it can also be applied to other scenarios where the transmitted signals are uncorrelated.