Study On Key Technologies Of Fusion Detection In Multisite Radar System

Multisite radar system(MSRS)has been a hot topic in the area of radar just since radar was invented.MSRS can illuminate a target from multiple viewing angles with widely separated local radar sites,which results in better performance compared to monostatic radar site.MSRS has attracted tremendous attention in engineering since it becomes more difficult for monostatic radar to improve its own performance.For engineering purpose,this dissertation studies some key technologies in the area of fusion detection for MSRS based on the previous works.The main contributions are shown as follows.1.Correlation analysis of target return in a diversity radar system.Based on previous works,we extended the signal model of MSRS from a plane coordinate system to a three-dimensional polar coordinate system.Under the assumption of narrow-band signals,the correlation coefficient of echoes from different diversity channels is calculated,where the target is modeled as infinite scatter points uniformly distributed in a three-dimensional spherical.The condition of statistical independence of the echoes from different diversity channels is analyzed.Besides,the theoretical results are verified by Monte Carlo simulations.2.Distributed detection algorithm based on censored data.A censoring scheme is studied in a scenario where the target only occupies a small fraction of resolution cells in the surveillance space.The local test statistics above local thresholds are transmitted to the fusion center,while those below the local thresholds are discarded.The optimal fusion rule is designed based on the Neyman-Pearson lemma.The statistical characteristics of local decisions are analyzed for MSRS with local radar sites receiving single or multiple returns.Global detection probability and false alarm rate are derived for performance analysis.The closed-form expressions of global false alarm rate or detection probability are obtained for the case where the observation follows the exponential distribution.Simulation results show that the method can approach a detection performance close to that of the centralized case with an extremely low communication rate.3.A communication bandwidth allocation algorithm based on the Ali-Silvey distance metric family.In some scenarios,the communication resource of fusion center in MSRS is limited.MSRS can achieve a better detection performance through proper communication bandwidth allocation.This dissertation uses two common distance metrics in the Ali-Silvey distance metric family,namely J-Divergence and the Bhattacharyya Distance,to address the issue that the optimal local decision criterion and the global decision criterion are coupled to each other.An optimization model for the communication bandwidth allocation algorithm is proposed based on the above criterion.The local optimal communication bandwidth allocation can be obtained by using the sequence quadratic programming(SQP)algorithm.The simulation results show that both metrics can effectively allocate the communication bandwidth to obtain a better detection performance than that of the average allocated bandwidth strategy.4.A low-communication rate fusion detection algorithm based on the generalized likelihood ratio criterion.In the search mode,exact signal-to-noise ratio(SNR)information of the target echo is often unpredictable,which leads to the difficulty of the Neyman-Pearson criterion to be directly used in engineering.It is assumed that each receiver can obtain multi-dimensional coherent signals,and the phases are statistically independent between different receivers.In this scenario,the well-known generalized likelihood ratio detection algorithm in the field of multidimensional signal processing is extended to MSRS.According to different observation scenarios,three low-communication rate fusion detection algorithms are designed to meet the communication bandwidth requirements.For the three fusion detection algorithms,the closed expressions of the false alarm probability are derived.The detection performance and communication rate requirements of the three algorithms are verified through theoretical analyses and numerical simulations.5.Fusion detection method based on the membership function.A robust fusion detection algorithm is designed for a scenario with large SNR divergence among different spatial diversity channels(SDCs).Inspired by the generalized likelihood ratio test(GLRT),the membership function is used to evaluate the maximum likelihood estimation of amplitude since exact estimates are difficult to achieve in the search stage.We substitute the evaluated estimation back into the likelihood ratio to derive the final fusion rule.Numerical simulations show that the proposed fusion detection algorithm can obtain better performance than the GLRT fusion and non-coherent accumulation fusion algorithm with large SNR divergence.At the same time,better robustness is achieved compared to existing algorithms.