A Research On Tracking-before-detection Algorithm Of Multiple Targets Based On The Distributed MIMO Radar

As a new field of radar, multiple-input multiple-output(MIMO) radar realizes the multi-channel and parallel transmission and acceptance of signal in space, by setting up a plurality of transmitting stations which generate different waveform signals(usually using quadrature signals) and a plurality of receiving stations which make a matched filter to the signal reflected by the targets and conduct a joint processing. Compared with conventional radar, MIMO radar can improve the performance of target detection and tracking in complex environments because of a higher resolving power to the target and more accurate estimation of angle. Among them, the distributed MIMO radar can obtain a larger gain of spatial diversity because of its special mode of setting station. However, the gain of spatial diversity at low SNR is not obvious. Therefore a possible way to improve the performance of MIMO-radar detection-and-tracking in low SNR environment is using the TBD technique. TBD technique makes a processing directly on the raw data with a long period of non-coherent integration, and gives the results of detection and tracking simultaneously.TBD technique is an effective means to deal with detection and tracking of the weak target.This thesis focuses on multi-target detection and tracking of distributed MIMO radar, mainly focusing on multi- target tracking algorithm which based on the theory of RFS and a TBD algorithm which based on the probability hypothesis density and adapted to the low SNR environment.Firstly, for the problem of multiple-target tracking under the circumstances that the number of targets is unknown or variable, this thesis introduces the theory of random sets, then focuses on the filtering algorithm of probability hypothesis density which based on the theory of RFS, and introduces two filtering realizations of probability hypothesis density: Gaussian mixture and particle filter.Secondly, considering the problem of detection and tracking for multiple targets under low SNR, this thesis focuses on the filter of probability hypothesis density which based on the framework of TBD, and optimizes the newborn mechanism of particles for the problems of the burden of calculation during in the implement of PF, low utilization rate of particles and poor filtering accuracy. Finally this thesis proposes an improved PHD-TBD algorithm which based on particle filter. The results of simulations show that the improved algorithm, to some extent, can improve the estimation performance of the target state.Finally, for the distributed MIMO radar, this thesis analyzes the signal and observation model of MIMO radar. Considering the problem of multi-sensor data fusion for MIMO radar, this thesis uses the strategy of joint-fusion filter, and proposes a multi-sensor TBD. The result of simulations shows that the algorithm is suitable for MIMO radar and improves the performance of target detection and tracking. For the problem that targets lost in the target-cross simulation, a strategy of clustering is adopted. Finally this thesis studies an improved detection and tracking algorithm which is applied to the closed and weak targets.