Detection of frequency hopped signals timing information using the temporal correlation function

Abstract

This work extends earlier work derived by Overdyk and investigates the use of wavelet transform and image processing tools to estimate hopping times occurring in frequency hopping schemes. The detection algorithm identifies frequency hopping time locations found in FH schemes from the information provided by the two-dimensional shortterm signal temporal correlation function. Hopping time locations are shown to be provided by indentifying TCF phase discontinuities. The detection scheme has three main stages: 1. Derive the analytic version of the FH signal and compute the resulting TCF function; 2. Enhance discontinuities via the one-dimensional Wavelet transform; 3. Apply morphological image processing operations and the Hough transform to estimate hopping time locations. Results show that for FH signals imbedded in additive White Gaussian noise, reliable detection performance may be obtained for SNR levels above 3 dB and good detection performance for SNR levels above 6dB for 5% to 20% detection accuracy.