Subspace Optimization in Centralized Noncoherent MIMO Radar


Distributed multiple input-multiple output (MIMO) radar is a research area that has received increasing attention lately, see, e.g., [5, 3, 10]. A salient feature of distributed radar is its ability to simultaneously engage a target from multiple aspect angles. While the legacy radar is confined to viewing the target from a single aspect angle at any given time instant, the distributed radar utilizes waveforms from spatially diverse stations to illuminate the target and detect reflected target energy from multiple aspect angles, taking advantage of aspect-dependent radar cross section (RCS) to significantly improve the ability to detect and track targets. The benefit over a single station implementation comes at the cost of increased system complexity including more demanding inter-station communications for data fusion and coordination among the stations. Given a multiplicity of stations, down-selecting the number of stations used in processing provides one mechanism to reduce system complexity, where the stations are selected in a manner that optimizes the performance for the number of resources that are dedicated to the task. We refer to this architecture as subspace centralized noncoherent MIMO radar (SC-MIMO). The SC-MIMO radar architecture, exemplified in Fig. 1, is characterized by the optimized selection of a subset of spatially diverse radar stations and joint processing of the received signals from this subset at a common fusion center. The transmitters are assumed to be sufficiently separated to yield spatially white channel transfer gains and are assumed to operate on a noninterference basis through time or frequency multiplexing, which facilitates both the separation of the signals at the receivers and the application of associated Doppler compensation tapering for signal conditioning. Subspace optimization measures in SC-MIMO are explored to optimize system performance in terms of probability of detection, information-theoretic capacity, and channel diversity, where optimized system performance in each of these senses is achieved by selecting the subspace that maximizes measures associated with the MIMO channel matrix. Information-theoretic metrics such as capacity and diversity are considered because of their ability to characterize MIMO channels in a manner that could potentially be exploited by an SC-MIMO system. For the case of SC-MIMO radar detection performance, joint detection optimization in a Neyman-Pearson (NP) sense with noncoherent square-law processing is shown to be equivalent to maximizing the Froebenius norm of the SC-MIMO radar channel matrix. The channel capacity measure is optimized by maximizing the determinant of the channel matrix [6, 12]. Diversity can be optimized by evaluating correlations between the elements of the channel matrix [8]. These subspace optimization measures are applied in the case of slowly changing channels wherein the

DOI: 10.1109/TAES.2011.5751254

6 Figures and Tables

Cite this paper

@article{Pratt2011SubspaceOI, title={Subspace Optimization in Centralized Noncoherent MIMO Radar}, author={Thomas G. Pratt and Yih-Fang Huang and Zhenhua Gong and Mike Lemmon}, journal={IEEE Trans. Aerospace and Electronic Systems}, year={2011}, volume={47}, pages={1230-1240} }