Search

The design of the waveform covariance matrix for beampattern matching in colocated multiple-input multiple-output (MIMO) radars represents a challenging problem because of its large number of variables and the presence of multiple constraints. The solutions available in the technical literature are computationally intensive and usually rely on iterative procedures that minimize a constrained mean square error (MSE). In this manuscript, a new computationally efficient method for beampattern matching design is proposed. This method, called sequential weight-shift unconstrained programming (SWSUP), allows to compute the covariance matrix of the probing signals achieving a desired beampattern at the transmit side of a colocated MIMO radar. Its derivation is based on the idea of reformulating the beampattern matching problem in an unconstrained form that can be tackled by breaking it into two subproblems. The first subproblem admits a closed-form solution, whose accuracy, in terms of MSE, is comparable to provided by other known methods. The solution of the second subproblem, instead, is evaluated through an iterative procedure and allows to achieve further improvement. Our numerical results evidence that the SWSUP method achieves precise beampattern matching with a substantially lower computational effort and computing time with respect to various existing alternatives.

It is well-known that the availability of transmit and receive arrays in colocated multiple-input multiple-output radar systems can be exploited to detect multiple targets, and estimate their range and angular coordinates. Unluckily, the accuracy achieved in the estimation of target parameters can be severely affected by hardware nonidealities. The impact of these nonidealities can be mitigated through the adoption of specific methods explicitly developed to estimate and compensate for them, i.e., briefly, of calibration methods. As far as we know, until now, most of the calibration techniques proposed in the technical literature for colocated multiple-input multiple-output radars are based on the idea of placing one or more reference targets in front of the considered radar device at perfectly known and fixed locations, and quantifying the difference between the expected radar image and the measured one. In this manuscript, two novel calibration methods are proposed. Both exploit the measurements captured by a colocated multiple-input multiple-output radar device at multiple positions; for this reason, the acquisition of such measurements involves the movement of the radar. The first one does not require any knowledge about the nature, number, position or radar cross section of the reference targets, provided that all these are in far-field; the second one, instead, needs a single reference target in near-field conditions, with the only constraints for it to be pointwise and isolated from stray targets in the range domain. This drastically simplifies the experimental setup to be employed for radar calibration. The proposed methods are tested on a commercial radar device and compared with a traditional method for radar calibration.

In this manuscript, the problem of detecting multiple targets and estimating their spatial coordinates (namely, their range and the direction of arrival of their electromagnetic echoes) in a colocated multiple-input multiple-output radar system operating in a static or slowly changing two-dimensional or three-dimensional propagation scenario is investigated. Various solutions, collectively called range & angle serial cancellation algorithms, are developed for both frequency modulated continuous wave radars and stepped frequency continuous wave radars. Moreover, specific technical problems experienced in their implementation are discussed. Finally, the accuracy achieved by these algorithms in the presence of multiple targets is assessed on the basis of both synthetically generated data and of the measurements acquired through three different multiple-input multiple-output radars and is compared with that provided by other methods based on multidimensional Fourier analysis and multiple signal classification.

Radars are expected to become the main sensors in various civilian applications, ranging from health-care monitoring to autonomous driving. Their success is mainly due to the availability of both low cost integrated devices, equipped with compact antenna arrays, and computationally efficient signal processing techniques. An increasingly important role in the field of radar signal processing is played by machine learning and deep learning techniques. Their use has been first taken into consideration in human gesture and motion recognition, and in various healthcare applications. More recently, their exploitation in object detection and localization has been also investigated. The research work accomplished in these areas has raised various technical problems that need to be carefully addressed before adopting the above mentioned techniques in real world radar systems. In this manuscript, a comprehensive overview of the machine learning and deep learning techniques currently being considered for their use in radar systems is provided. Moreover, some relevant open problems and current trends in this research area are analysed. Finally, various numerical results, based on both synthetically generated and experimental datasets, and referring to two different applications are illustrated. These allow readers to assess the efficacy of specific methods and to compare them in terms of accuracy and computational effort.