Array Position Optimisation for Compressed Sensing MIMO Radar based on Mutual Coherence Minimisation

Abstract

In this paper, an optimization methodology for repositioning antenna elements of a collocated Compressed Sensing (CS) based Multiple Input Multiple Output (MIMO) radar, to improve target detection performance, by minimizing the mutual coherence of the associated sensing matrix has been suggested. We initialize the problem as a mutual coherence of the sensing matrix resulting from a simple 3Tx/4Rx Uniform Linear Array (ULA) restricted by an array aperture of specified size, and then reposition the elements within the restricted aperture such that the value of mutual coherence reduces. The optimization problem is formulated as minimizing the l∞ norm of the Gramian of the associated sensing matrix, the global optimization solver simulated annealing is considered to solve the nonconvex problem. The optimized array's performance is evaluated against a ULA, Co-prime array, and Sparse array by comparing metrics such as the probability of perfect reconstruction (Recovery percentage) and Recovery error (root mean square error (rmse)) for scenes with multiple targets and different SNR values, using Monte Carlo simulations. The study demonstrates the methodology to generate a random array, which results in low mutual coherence of its respective sensing matrix, which consequently results in improved performance of the CS-MIMO radar.

Sequence optimisation for compressed sensing CDMA MIMO radar via mutual coherence minimisation

“…Additionally, this projection matrix (if designed correctly) can serve as a means to exploit signal sparsity, in addition to the conventional scene sparsity imposed by basis transformation matrices [32]. A demonstration of this concept was published in our earlier study [52], where we designed a sparse random array based on MC minimisation. Building upon the concepts of our current and past studies, our future work will focus on defining MC as a design metric for sparsity-aware radar systems, which will be achieved through the careful design of a sparsifying randomised projection matrix.…”

“…Building on our initial analysis in ref. [31], we explored the influence of different code sequences on CS reconstruction performance, laying the groundwork for this article.…”

Elevation Profile Estimation For Single Pass Bi-Static SAR Tomography Using Compressed Sensing

“…Based on the choice of d T the receiver array can be configured. An extensive study on the types of array configurations can be found here [6].…”

“…Figure 2b demonstrates that spectral-based methods and CS-based methods yield equivalent results when the targets are separated by δh = 0.8m. However, in Figure 2d, spectral methods fail to resolve or accurately detect the targets, while CS-based methods exhibit greater resilience in such cases matrix has to satisfy the Restricted Isometry property (RIP) or should be designed to have low mutual coherence, detailed information on this topic can be found in [6]. In the context of this study, we chose an arbitrary array constellation, which results in a lower mutual coherence than a regular ULA configuration.…”

Randomised Compression Ratios for Effective Large Point Cloud Processing Using Compressive Sensing

“…The proposed method aims to integrate the core principles of CS with the foundational concepts of established data compression methodologies. It leverages the benefits of randomised projections, a technique that has shown promise in efficiently encoding high-dimensional data into a lower-dimensional space without significant loss of information (Nagesh et al, 2022). This integration aims to preserve the precision in scene interpretation, a critical aspect often compromised in traditional methods, while optimising computational efficiency.…”