PERFORMANCE ANALYSIS ON A DECOUPLED MAXIMUM LIKELIHOOD ANGLE ESTIMATOR IN AN FM BASED PASSIVE BISTATIC RADAR Geun-Ho Park, Dong-Gyu Kim, Ho Jae Kim, Xianglan Jin, Jin-Oh Park * , Won-Jin Lee * , Jae Heon Ko * , Hyoung-Nam Kim © Pusan National University, LIG Nex1 * Busan/Republic of Korea, Seongnam/Republic of Korea * hnkim@pusan.ac.kr © ABSTRACT We analyze the performance of a decoupled maximum likelihood (DEML) angle estimator in an FM based passive bistatic radar. Under the assumption that the bistatic range, the Doppler frequency estimates, and the transmitted signal emitted from an illuminator are given, we analyze the sensitivity of the DEML estimator against the range and Doppler frequency errors from the analytic expressions and simulation results. As a result, the performance of the DEML estimator may be mainly degraded by the Doppler frequency errors and we derive a condition that the root-mean-square error of the DEML estimator diverges. KEY WORDS Decoupled maximum likelihood angle estimator, FM based bistatic radar. 1. Introduction A passive bistatic radar exploits the non-radar transmitters of opportunity, such as FM (frequency modulated) radio, DVB (digital video broadcasting), DAB (digital audio broadcasting), and GSM (global system for mobile communications) for detecting the multiple fast moving targets [1-2, 8-11]. In order to detect K targets, the relative bistatic range, Doppler frequency, and angle- of-arrival estimates corresponding to each target should be extracted from the received signals. The angle-of- arrival estimates can be extracted from an antenna array with M sensors. The maximum likelihood approach may be considered to estimate the parameters [3]. Since the maximum likelihood approach performs a K-dimensional search [3], however, the computational burden in this search problem is dramatically increased according to the number of targets. To cope with this problem, a decoupled maximum likelihood (DEML) angle estimator [4] was proposed to improve the computational complexity of the maximum likelihood technique in [3] by using an assumption that a priori information of the multiple source signals are given; that is, the source signals are assumed to be known. Thus, the DEML estimator can be easily applied to the communication systems by using the preamble information. After that, several joint estimation techniques [5-7] which can estimate the range, the Doppler shift, and the angle-of-arrival, simultaneously, were suggested. However, since the range and the Doppler frequency values are given in some applications, these joint estimation algorithms may not be effective for such applications as the passive radar systems because of the computational burden of these joint estimation techniques. Since the DEML angle estimator estimates only angle values, we have focused on the DEML estimator to utilize the simplicity of the DEML angle estimator. When we consider applying the DEML angle estimator in passive radar areas, however, it is difficult to use a priori information of incident signals directly. Since the target echo signals are time-delayed and Doppler frequency shifted versions of the transmitted signal, the source signals should be obtained from the estimates of the relative bistatic range and Doppler frequency. In case of using the estimates of the source signals, the relative bistatic range and Doppler frequency errors may cause a critical problem in the DEML angle estimator. In this paper, we investigate the feasibility of the DEML angle estimator when we have the range and Doppler frequency estimates with the critical errors. To figure out the sensitivity of the DEML estimator against the range and Doppler frequency errors, we analyze the effect of the range and Doppler frequency errors from several analytic expressions and simulation results. This paper is organized as follows. Section 2 presents the signal models of a received signals and defines the problem of interest. In section 3, the DEML angle estimator is briefly reviewed, and the performance analysis of the DEML angle estimator is covered in section 4. Simulation results are presented in section 5 and conclusions are drawn in section 6. 2. Problem Formulation Consider the angle estimation of K narrowband target echoes s k (t) (k = 1,…, K) impinging on an arbitrary array with M sensors. Under the assumption that the interferences are eliminated, then a received signal x(t) is given by 1 () , ) () ( ), K k k k k t s t t x a n        (1) where a(θ k , ϕ k ) denotes the steering vector corresponding to the elevation θ k and azimuth angle ϕ k of the kth target echo signal, and n(t) is a noise vector of the white Proceedings of the IASTED International Conference February 20 - 21, 2017 Innsbruck, Austria Modelling, Identification and Control (MIC 2017) DOI: 10.2316/P.2017.848-044 242