Detection of spread-spectrum signals using the time-domain filtered cross spectral density

Abstract

The authors present a technique for the detection of spread-spectrum signals, of arbitrary form, even when the signal power spectral density (PSD) is well below the surveillance receiver noise spectral density, using a pair of antennas with broadband (1 GHz or more) receivers. Cross correlating the outputs of two receivers, spatially separated by a distance of the order of one metre or more, produces a cross-correlation function (CCF), in which the noise components are spread uniformly over the whole width, while the signal component, the narrow autocorrelation function (ACF) of the spread-spectrum signal, is concentrated in the centre. A simple time-domain filter can select this narrow centre portion of the CCF, rejecting the remainder which contains only noise. Taking the Fourier transform of this windowed CCF produces the 'time-domain filtered cross spectral density' (TDFCSD), in which the signal to noise ratio is independent of receiver bandwidth. Spread-spectrum signals can then be both detected and characterised, in an extremely sensitive broadband system, by threshold detection in this TDFCSD. An analysis of the performance of this dual-receiver system is presented. A computer simulation illustrates the signal processing process involved, and shows excellent agreement with the analysis. A time-integrating acousto-optic correlator may be used to perform the cross correlation and time domain filtering of wideband signals in real time. Final processing of the much reduced data set to obtain the TDFCSD can be carried out digitally.