非一致噪声下的多阵列直接定位算法

A Direct Position Determination Algorithm Based on Multi Arrays in the Presence of Unknown Nonuniform Noise

针对实际无源定位系统中多个阵列分散分布以及单个阵列各阵元之间差异造成的噪声功率分布不均匀导致直接定位(DPD)算法精度下降的问题,提出了一种非一致高斯白噪声场中的多阵列最大似然DPD(UN-ML-DPD)算法,并推导出该条件下的克拉美罗界。首先计算各阵列输出协方差矩阵;再将计算结果传到数据处理中心;然后由数据中心通过迭代的方式同时对目标位置与非一致噪声功率进行最大似然联合估计,以收敛后得到的结果作为目标位置精确估计值,从而减弱了非一致噪声对DPD算法的影响;最后在多目标条件下用交替投影法降低算法复杂度。较之传统算法,UN-ML-DPD算法能够提高多阵列在低信噪比下的定位精度。仿真结果表明:UN-ML-DPD算法在-15dB的低信噪比下估计误差小于15km,与DPD算法相比定位精度提高15%以上;能较为准确地估计各阵列噪声协方差矩阵,在标准噪声功率小于30 W时估计误差小于3.5 W;在高信噪比下定位精度能够逼近克拉美罗下界。

A multi-array based maximum likelihood direct position determination （DPD） algorithm in the presence of unknown nonuniform noise （UN-ML-DPD） is proposed and an expression of Cramér-Rao lower bound （CRLB） is derived to solve the problem that the position accuracy of the DPD algorithm decreases due to dispersed multi-array distribution and nonuniform noise power distribution caused by difference among array elements in passive localization systems. The covariance matrixes of the array data are calculated and the results are transmitted to the process center. Then, the power of the nonuniform noise and the position of the targets are co-estimated simultaneously using the maximum likelihood method with iteration, and the converged result is taken as the target position estimation, so that the effect of the nonuniform noise is reduced. The alternative projection method is utilized to lower the algorithm complexity when multi targets exist. Compared with the traditional method, the UN-ML-DPD algorithm can improve the position accuracy under low signal-to-noise ratio （SNR）. Simulation results show that the estimation error is lower than 15 km for SNR is --15 dB, and the accuracy of the position estimation is increased by 15% compared with the DPD method. The noise covariance matrix is well estimated, and the estimation error is less than 3.5 W when the standard power of the noise is lower than 30 W. Moreover, the position accuracy approaches the CRLB when the SNR are high.