基于复数因子分析模型的步进频数据压缩感知

摘要

It usually takes a long observing timewhen a cognitive radar transmits the High-Range-Resolution (HRR) stepped-frequency signal. To save time, partial pulses of the stepped-frequency signal are transmitted to obtain the incomplete frequency data, and a Bayesian reconstruction algorithm is proposed to reconstruct the corresponding full-band frequency data. Firstly, the Complex Beta Process Factor Analysis (CBPFA) model is utilized to statistically model a set of full-band frequency data, whose probability density function (pdf) can be learned from this CBPFA model. Secondly, when the target is tracked and its attitude changes not much, the cognitive radar can just transmit the partial pulses of the stepped-frequency signal, and the corresponding full-band frequency data can be analytically reconstructed from the incomplete frequency data via the Compressive Sensing (CS) method and Bayesian criterion based on the previous pdf learned with CBPFA model. The reconstruction experiments of the measured HRR data demonstrate the performance of the proposed method.%认知雷达发射高距离分辨率步进频信号通常需要较长的观测时间。为了节省时间资源，该文提出一种贝叶斯重构算法，用较少的步进频信号脉冲得到的频点缺失频域数据，重构出相应的全带宽频域数据。首先利用复数贝塔过程因子分析(Complex Beta Process Factor Analysis, CBPFA)模型对一组全带宽频域数据进行统计建模，求解得到其概率密度函数；然后在目标被跟踪且姿态变化不大的情况下，只发射步进频信号的部分脉冲，根据先前CBPFA模型得到的概率密度函数，对频点缺失的频域数据利用压缩感知理论和贝叶斯准则解析地重构出相应的全带宽频域数据。基于实测1维高分辨距离(High Range Resolution, HRR)数据的重构实验，证明了该文提出方法的性能。