OFDM passive radar employing compressive processing in MIMO configurations

Abstract

A key advantage of passive radar is that it provides a means of performing position detection and tracking without the need for transmission of energy pulses. In this respect, passive radar systems utilising (receiving) orthogonal frequency division multiplexing (OFDM) communications signals from transmitters using OFDM standards such as long term evolution (LTE), WiMax or WiFi, are considered. Receiving a stronger reference signal for the matched filtering, detecting a lower target signature is one of the challenges in the passive radar. Impinging at the receiver, the OFDM waveforms supply two-dimensional virtual uniform rectangul ararray with the first and second dimensions refer to time delays and Doppler frequencies respectively. A subspace method, multiple signals classification (MUSIC) algorithm, demonstrated the signal extraction using multiple time samples. Apply normal measurements, this problem requires high computational resources regarding the number of OFDM subcarriers. For sub-Nyquist sampling, compressive sensing (CS) becomes attractive. A single snap shot measurement can be applied with Basis Pursuit (BP), whereas l1-singular value decomposition (l1-SVD) is applied for the multiple snapshots. Employing multiple transmitters, the diversity in the detection process can be achieved. While a passive means of attaining three-dimensional large-set measurements is provided by co-located receivers, there is a significant computational burden in terms of the on-line analysis of such data sets. In this thesis, the passive radar problem is presented as a mathematically sparse problem and interesting solutions, BP and l1-SVD as well as Bayesian compressive sensing, fast-Besselk, are considered. To increase the possibility of target signal detection, beamforming in the compressive domain is also introduced with the application of conve xoptimization and subspace orthogonality. An interference study is also another problem when reconstructing the target signal. The networks of passive radars are employed using stochastic geometry in order to understand the characteristics of interference, and the effect of signal to interference plus noise ratio (SINR). The results demonstrate the outstanding performance of l1-SVD over MUSIC when employing multiple snapshots. The single snapshot problem along with fast-BesselK multiple-input multiple-output configuration can be solved using fast-BesselK and this allows the compressive beamforming for detection capability.