Research On CS-based Robust Broadband DOA Estimation Method

Array signal processing refers to spatial sampling and extracting target feature via array,which is composed of several spatially placed sensors.As time-delay exists between signals' arriving at these sensors,space parameters could be determined once signal spectrum is obtained by analyzing the phase difference of receipt signals.Normally,DOA(direction-ofarrivals)estimation is based on the assumptions of narrow-band scenario,which does not fit the situation in shallow water as signals generally propagate in broadband form.Array errors are not taken into account either,while the robustness of sensors' receiving characteristics could not be guaranteed.Broadband signals precede narrow-band signals in terms of resolution performance and information capacity.Hence broadband system is of great significance in research of DOA estimation for its high detection precision and robustness to interferences.Estimation algorithms at present are not able to realize satisfying resolution and robustness against noise in ocean where sampling costs a lot.Known in recent years for its unique perspective of sampling,compressive sensing makes it possible to reconstruct signals without distortion on limited sampling condition.DOA is a key parameter to realize source separation in linear array.However,acoustic signal in shallow water would be reflected and refracted by the surface and bottom of sea,which makes it hard to separate raypaths in certain situations.Double-array structure represents signals from three dimensions so that the raypath could be uniquely identified.Nevertheless,System errors exist in the sensor array on account of its inherent defects,resulting in severe performance degradation in DOA estimation algorithms.These errors exist in this structure and have a great impact on the performance of the separation algorithms as well.In addition,error condition gets more complicated than that in linear array as both transmitting array and receiving array should be considered.To improve the performance of broadband DOA estimation methods,the algorithm proposed in this thesis is based on compressive sensing and enhances resolving ability as well as robustness against noise.Still,the wideband DOA estimation method proposed in this thesis is not able to overcome the array uncertainty either.In order to improve the robustness of the wideband DOA estimation method against array errors,this thesis provides a wideband-based calibration method,which diminishes the impact of the array errors effectively.To enhance the robustness of DOA estimation algorithm against errors in highly dimensional arrays,especially for double-arrays,joint error model is needed to accomplish the calibration in a unified framework.This thesis proposes a joint error calibration method applied in double-array architecture.Simulation results and real underwater experiments verified the effectiveness of the methods.