Bootstrap-based detection of signals with unknown parameters in unspecified correlated interference

A detection scheme that uses estimators for signal frequency, amplitude, and phase to detect signals in correlated interference is proposed. The scheme does not require training data but uses the bootstrap to estimate the distribution of the test statistic under the hypothesis directly from the received data. Simulation results are provided that confirm the detector performance for correlated Gaussian and non-Gaussian interference and the consistency of the estimators. The detector asymptotically maintains the preset level of false alarm, and a detection rate of 100% is achievable at -5 dB signal-to-noise ratio (SNR) for sample sizes as small as 500 data points. The case where the arrival time of the signal is unknown is also treated. Simulation results showing the performance of the detector, when it is suitably extended to handle this case, are given.