TY - JOUR

T1 - Maximum Likelihood Based Sensor Array Signal Processing in the Beamspace Domain for Low Angle Radar Tracking

AU - Zoltowski, Michael D.

AU - Lee, Ta-Sung

PY - 1991/1/1

Y1 - 1991/1/1

N2 - A 3-D beamspace domain, maximum likelihood (3-D BDML) bearing estimation scheme for low angle radar tracking is developed. The novelty of 3-D BDML is its judicious exploitation of the fact that the respective beams associated with any three classical beam-forming vectors which are mutually orthogonal have M - 3 nulls in common, where M is the number of elements comprising a uniformly spaced, linear array. Exploitation of this property yields an estimation scheme that is nearly as computationally simple as the 2-D beamspace bearing estimation technique based on sum and difference beams employed in conventional monopulse radar. To provide robustness to the severe signal cancellation occurring across the array when the direct and specular path signals arrive 180° out of phase at the aperture center with roughly equal amplitude, frequency diversity is incorporated into 3-D BDML. The coherent signal subspace concept of Wang and Kaveh is invoked as a means of retaining the computational simplicity of single frequency 3-D BDML. It is shown that if the auxiliary frequencies are chosen from a restricted set of “special” values, the 3 x 3 beamspace domain based focusing matrices necessary for coherently combining the target energy at each transmission frequency do not depend on the bearings of the direct and specular signals and are known a priori. Under these conditions, perfect “focusing” may be achieved at the outset, i.e., without iterating, such that the computational complexity is essentially that associated with single frequency operation.

AB - A 3-D beamspace domain, maximum likelihood (3-D BDML) bearing estimation scheme for low angle radar tracking is developed. The novelty of 3-D BDML is its judicious exploitation of the fact that the respective beams associated with any three classical beam-forming vectors which are mutually orthogonal have M - 3 nulls in common, where M is the number of elements comprising a uniformly spaced, linear array. Exploitation of this property yields an estimation scheme that is nearly as computationally simple as the 2-D beamspace bearing estimation technique based on sum and difference beams employed in conventional monopulse radar. To provide robustness to the severe signal cancellation occurring across the array when the direct and specular path signals arrive 180° out of phase at the aperture center with roughly equal amplitude, frequency diversity is incorporated into 3-D BDML. The coherent signal subspace concept of Wang and Kaveh is invoked as a means of retaining the computational simplicity of single frequency 3-D BDML. It is shown that if the auxiliary frequencies are chosen from a restricted set of “special” values, the 3 x 3 beamspace domain based focusing matrices necessary for coherently combining the target energy at each transmission frequency do not depend on the bearings of the direct and specular signals and are known a priori. Under these conditions, perfect “focusing” may be achieved at the outset, i.e., without iterating, such that the computational complexity is essentially that associated with single frequency operation.

UR - http://www.scopus.com/inward/record.url?scp=0026119833&partnerID=8YFLogxK

U2 - 10.1109/78.80885

DO - 10.1109/78.80885

M3 - Article

AN - SCOPUS:0026119833

VL - 39

SP - 656

EP - 671

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 3

ER -