The Algorithm And System Of Time Difference Of Arrival Based Passive Acoustic Target Localization

Acoustic array network has been widely used in military applications, including monitoring vehicles and helicopters in the battlefield at low altitude. Compared with radar, acoustic array network is a passive detection tool and it has features of small size, good invisibility, strong anti-interference, low power, low cost and a wide range of detecting field. So it has great potential value in the future. This paper presents solutions in solving the problems of high accuracy and low power consumption of target location using time difference of arrival. Main contributions are summarized as follows:1. This paper aims to obtain the signal TDOA between two acoustic array nodes, which mean-s range difference in multi-target localization scene. We propose three TDOA algorithm-s:Eigenvalue Decomposition Based Maximum Likehood,(EDBML), Compressive Sensing Based Cross Correlation(CSBCC), and Enhanced Compressive Sensing Based Cross Cor-relation(ECSBCC). EDBML algorithm can get high precision estimation of TDOA. This method should be applied in nodes with powerful calculation capacity and sufficient ener-gy supply. For limited communications, we propose CSBCC method to estimate TDOA, by which means we are able to determine TDOA by few acoustic signals, thereby greatly reducing acoustic node power consumption. Then we optimize measurement matrix in the compression process in order to significantly improving the precision of CSBCC estimation. Simulation results show that the EDBML algorithm can obtain high accuracy TDOA and CSBCC algorithm is able to maintain stable and low error in the case of signal was greatly compressed. And ECSBCC has a much better performance of CSBCC algorithm.2. This paper presents three Time Difference of Arrival(TDOA) based localization algorithm-s:Circle Center Based Least Square Estimator(CCB-LSE), Circle Center Based Total Least Square Estimator(CCB-TLSE) and Sensing Probability Based Multi-Target Localization(SPBML). In Single target scenario, the CCB-LSE algorithm establishes linear least squares to obtain closed-loop solutions of unknown locations. In order to improve the accuracy of CCB-LSE algorithm, we use singular value decomposition to weaken the noise and establish the CCB-TLSE. In multiply target localization scene, the features of multiple target signals are quite similar and we have no prior information.In order to solve the problem, we study SPBML method. Finally, the simulation results show that the CCB-LSE and CCB-TLSE are able to avoid the divergence problem in CCB-MLE and remain very low computational complexity. The accuracy of CCB-TLSE is almost as minimum as CRLB. SPBML algorithm has a high probability of eliminating ghost points.3. For verifying the reliability of CCB-TLSE, SPBML, EDBML, and ECSBCC algorithms, this paper independently presents a TDOA based acoustic array localization system. This system consists of multiple eight-layer PCB board based microphone array nodes, a sink node, target source and a powerful fusion center. This system has two working modes:complete signal mode and incomplete signal mode. The complete signal mode and incomplete signal mode use EDBML and ECSBCC method to determine TDOA, respectively. In order to verify the proposed localization system, we conduct TDOA measurement and localization experiments in the two modes. Experimental results showed that acoustic array nodes are running well, the system can work perfectly in real-world scenario, TDOA measurements and target local-ization results can meet expectations, indicating that the proposed acoustic array localization system can work well.