DOA Estimation of Low Altitude Target Based on Bounded Component Analysis Algorithm Eren Babatas 1,2 , Alper T. Erdogan 1 [1] Koc University, Istanbul, Turkey [2] Aselsan Inc., Ankara, Turkey Keywords: Blind Source Separation, Bounded Component Analysis, Direction Of Arrival Estimation Abstract A recently introduced Blind Source Separation method, called Bounded Component Analysis, is used as preliminary tech- nique to isolate direct path radar wave from ground reflected waves in order to overcome the multipath effect. This method enables the radar to estimate the target angle without any a priori knowledge of the operation environment. The numeri- cal experiments illustrate the potential benefit of the proposed approach relative to classical maximum likelihood method (CMLM) based on free space propagation model. 1 Introduction Direction Of Arrival (DOA) estimation of a low altitude tar- get is still a current issue that engages researchers’ attention. While estimating the DOA of low altitude target, radar re- ceive two or more coherent echoes via multipath. This re- sults the received signals to be faded and strengthened cycli- cally while target propagation. Many researchers have inves- tigated the multipath effects on radar and suggested various approaches to reduce the estimation errors in DOA estima- tion brought by these effects. The methods based on these ap- proaches can be classified in three main categories: Monopulse methods, parametric methods and subspace methods. The clas- sical maximum likelihood method (CMLM) [1] is classified as a parametric method and can distinguish coherent signals. The CMLM solves the maximum likelihood function in order to maximize the correlation between array manifold matrix and received data. However, the CMLM does not show a good per- formance if the direct-path and reflected-path echoes, which are coherent signals, fade each other. In this paper, a Blind Source Separation (BSS) technique named Bounded Component Analysis (BCA) is taken into con- sideration to separate the direct-path and reflected-path echoes which can improve the performance of the CMLM in the fad- ing case. BCA is a recently introduced BSS scheme that uti- lizes the boundedness property of sources to replace the usual independence assumption with a weaker assumption, allowing separation of both dependent and independent sources [2]. In [3], a geometric BCA framework is introduced which can sep- arate both independent and dependent sources from their mix- tures where the mixing system is instantaneous. This approach is based on the assumption that bounded source vectors lie in a rectangular l ∞ norm ball. Since radar echoes are bounded by Effective Radiation Power (ERP), this BCA framework is ap- plicable to the multipath problem. We combine this approach with the CMLM [1] to solve the low angle problem. The pro- posed algorithm weakens the influence of the multipath atten- uation and improves the performance of low-angle estimation. 2 Multipath Signal Model Griesser and Balanis’ model [4] is taken as the multipath signal model. Assumed geometry by the model is given in fig. 1 below. As shown in fig. 1, a uniform linear array (ULA) and a Fig. 1: Multipath signal model for radar wave propagation. point target are utilized. In transmitting and receiving, there are two ways between the antenna and the target, i.e., direct-way (r d,tr ,r d,m ) and reflection-way (r 1,tr + r 2,tr ,r 1,m + r 2,m ). Reflection angles are represented with ψ tr ,ψ m . The earth is modelled as a sphere with radius R e . The reflection coefficient of the spherical earth, ρ(n,ψ), is a complex variable. For the geometrical calculation of the reflection angles ψ tr ,ψ m , we use the approximate formulas given in [4]. The reflection coefficient ρ(n,ψ) is the sum of two components called specular reflection coefficient (ρ s ) and diffuse reflection coefficient (ρ d ). ρ(n,ψ)= ρ s (ψ)+ ρ d (n) (1) The specular reflection coefficient is calculated with the subse- quent equations: 1