A space-time estimator for the detection and estimation of multiple sinusoidal signals

Detection and estimation of sinusoids is an important task in many applications such as audio analysis, radar and sonar. Most existing estimation techniques use a single discriminant, such as either frequency or direction-of-arrival and are limited by a one-dimensional resolution limit and a one-dimensional improvement in SNR. Instead, we propose a novel space-time estimator for the detection, estimation and super-resolution of multiple signals using an adaptive iterative analysis technique. In our method, each component is detected and adaptively cancelled in turn to eliminate bias caused by sidelobe leakage. This technique is based on the incremental multi-parameter algorithm (IMP), modified to estimate both the directions and frequencies of narrowband components in data from a wideband sensor array. The basic resolution is two-dimensional using a simultaneous two-sided matched filter transform to enhance the SNR and provide better accuracy and detection probability. The method can be extended to include tracking in two dimensions.