Research On DOA Estimation Methods For MIMO Radar Based On Channel Compression

In modern radar and communication technology,the location information of targets(or users)is of great significance in the whole system.Hence,high-accuracy direction finding systems and algorithms have gained sufficient research currently.However,in almost all these systems,there is a common problem that the number of front-end circuit chains is too high.Thus,a serious of problems emerges,such as decreasing on channel isolation,increasing on channel uncertainty,and rising of system cost and computational burden.Therefore,it is in need of reducing the number of front-end circuit chains while guaranteeing the estimation accuracy.The widespread multiple-input multiple-output(MIMO)radar is taken as the research objects to obtain low-complexity direction finding structure with high estimation performance,where the compressed sensing technique is used to reduce the system complexity.First,aiming at the high signal dimension in MIMO radar,this thesis proposes the compressive MIMO radar structure.A compression matrix is inserted after the matched filter bank in the proposed structure.The signals are therefore linearly combined,and projected to a low-dimensional space.Consequently,the signal dimension is reduced.At the same time,to reveal the estimation performance of the proposed structure,the Cramer-Rao bound(CRB)for DOA estimation using the proposed compressive MIMO is derived,thus providing a theoretical guidance for the estimation performance of the proposed structure.In addition,a randomly generated compression matrix is usually used.However,the use of random compression matrix will lead to a relatively high information loss.To address this issue,a maximum mutual information criterion is proposed,where the prior information of the incident directions of signal is utilized.Thus,the estimation accuracy is improved by using the optimized compression matrix.According to the simulation validation,the complexity is effectively reduced with the proposed structure.Furthermore,when compared to conventional MIMO radar with the same signal dimension,the proposed structure has a higher estimation accuracy.Then,to reduce the number of required transmit and receive antenna in MIMO radar,this thesis makes further research on sparse MIMO radar.The corresponding research can be divided into two parts.In the first part,the hole problem existing in the sum-difference coarray of sparse MIMO radar is considered.Similar as sparse array,the holes lead to aperture loss,thus compromising the estimation performance.The matrix completion method is extended to the sparse MIMO radar model in this thesis,where a nuclear norm-based optimization problem is proposed.Through searching the optimum solution of the proposed problem,the missing lags is interpolated using the received data.Thus,the lags that are out of the continuous range are utilized,and the estimation accuracy and DOF are improved.In the second part,that problem that the number of channels is too many after matched filters is taken into account.To address this issue,the same method that is involved in compressive MIMO radar is used,where the compressive array is exploited in the sparse MIMO radar.Then,the compressive sparse MIMO radar structure is proposed in this thesis for DOA estimation.The proposed structure added a compression matrix after the matched filters.As such,the received signals are linearly combined,and the number of channels is reduced.Hence,the system complexity is reduced.At the same time,the signal dimension is also decreased,thus releasing the computational burden.Finally,this thesis makes further improvements on compressive sparse MIMO radar,and proposes a compressive MIMO radar with multiple carrier frequencies.The proposed structure transmits multiple groups of signals with different carrier frequency for detection,leading to an increasement on the target information.The estimation performance is then improved without changing the system configuration.To fully utilize the information embedded in all the signals,a DOA estimation algorithm based on group sparsity reconstruction is proposed in this thesis.In the proposed algorithm,the cross-correlation information between signals associated with different carrier frequencies is used to extend the degree of freedom,while the auto-correlation information is taken into account to suppress the spurious peaks and improve the estimation accuracy.In addition,the corresponding CRB is derived,where the existence conditions are also analyzed.Thus,a theoretical foundation is provided for the estimation accuracy and degree of freedom that can be obtained by the proposed structure.By conducting multiple groups of simulation experiments,the superiorities of the proposed structure is verified,where the required number of physical transmit and receive antennas is reduced while remaining the estimation performance unchanged.