Showing papers by "Alexander M. Haimovich published in 2019"

End-to-end Learning of Waveform Generation and Detection for Radar Systems

Abstract: An end-to-end learning approach is proposed for the joint design of transmitted waveform and detector in a radar system. Detector and transmitted waveform are trained alternately: For a fixed transmitted waveform, the detector is trained using supervised learning so as to approximate the Neyman-Pearson detector; and for a fixed detector, the transmitted waveform is trained using reinforcement learning based on feedback from the receiver. No prior knowledge is assumed about the target and clutter models. Both transmitter and receiver are implemented as feedforward neural networks. Numerical results show that the proposed end-to-end learning approach is able to obtain a more robust radar performance in clutter and colored noise of arbitrary probability density functions as compared to conventional methods, and to successfully adapt the transmitted waveform to environmental conditions.

Waveform Optimization in Cloud Radar with Spectral Congestion Constraints

Abstract: We investigate a cloud radar system consisting of a radar transmitter and distributed nodes linked to a remote processing center (PC) via multiple-access wireless backhaul channels. Each node serves both as a radar receiver and relay that forwards to the PC an amplified version of the received signal. To accommodate the ever-growing demand for spectrum, the cloud radar system is required to operate over a spectrum partially shared with communication devices and services. We formulate a problem of jointly optimizing the radar waveform as well as the relay power gains to maximize a mutual information criterion that serves as proxy for detection performance. The optimization requires knowledge of second-order statistics of clutter and interference, and is constrained by limitations imposed by the bandwidth shared with communication. Numerical results show that the joint optimization is advantageous over optimizing only the radar waveform or only the relay gains.

Blind Sparse Estimation of Intermittent Sources Over Unknown Fading Channels

Abstract: Radio frequency sources are observed at a fusion center via sensor measurements made over slow flat-fading channels. The number of sources may be larger than the number of sensors, but their activity is sparse and intermittent with bursty transmission patterns. To account for this, sources are modeled as hidden Markov models with known or unknown parameters. The problem of blind source estimation in the absence of channel state information is tackled via a novel algorithm, consisting of a dictionary learning (DL) stage and a per-source stochastic filtering (PSF) stage. The two stages work in tandem, with the latter operating on the output produced by the former. Both stages are designed so as to account for the sparsity and memory of the sources. To this end, smooth LASSO is integrated with DL, while the forward-backward algorithm and Expectation Maximization (EM) algorithm are leveraged for PSF. It is shown that the proposed algorithm can enhance the detection and the estimation performance of the sources, and that it is robust to the sparsity level.

Blind Sparse Estimation of Intermittent Sources over Unknown Fading Channels

Abstract: Radio frequency sources are observed at a fusion center via sensor measurements made over slow flat-fading channels. The number of sources may be larger than the number of sensors, but their activity is sparse and intermittent with bursty transmission patterns. To account for this, sources are modeled as hidden Markov models with known or unknown parameters. The problem of blind source estimation in the absence of channel state information is tackled via a novel algorithm, consisting of a dictionary learning (DL) stage and a per-source stochastic filtering (PSF) stage. The two stages work in tandem, with the latter operating on the output produced by the former. Both stages are designed so as to account for the sparsity and memory of the sources. To this end, smooth LASSO is integrated with DL, while the forward-backward algorithm and Expectation Maximization (EM) algorithm are leveraged for PSF. It is shown that the proposed algorithm can enhance the detection and the estimation performance of the sources, and that it is robust to the sparsity level.

Effects of System Impairments on the Performance of Distributed STAP

Abstract: In this paper, we study the effect of phase noise and other impairments on space-time adaptive processing. Performance of distributed space-time adaptive processing (STAP) with respect to system aspects and impairments specific to a coherent distributed system is also analyzed. The combined effects on STAP performance stemming from the distributed architecture: frequency offsets, generalized motion, sparse aperture, and sensor location errors, were also analyzed. Simulation results illustrate the effect of individual impairment and combined errors on degrading the performance of target detection.

Sparse Dictionary Learning and Per-source Filtering for Blind Radio Source Separation

Abstract: Radio frequency sources are observed at a fusion center via sensor measurements made over slow unknown flat-fading channels. The number of sources may be larger than the number of sensors, but their activity is sparse and intermittent with bursty transmission patterns accounted by hidden Markov models. The problem of blind source estimation in the absence of channel state information is tackled via a novel algorithm, consisting of a dictionary learning (DL) stage and a per-source stochastic filtering (PSF) stage. To this end, smooth LASSO is integrated with DL, while the forward-backward algorithm is leveraged for PSF. It is shown that the proposed algorithm can enhance the detection performance of the sources.

End-to-end Learning of Waveform Generation and Detection for Radar Systems

Abstract: An end-to-end learning approach is proposed for the joint design of transmitted waveform and detector in a radar system. Detector and transmitted waveform are trained alternately: For a fixed transmitted waveform, the detector is trained using supervised learning so as to approximate the Neyman-Pearson detector; and for a fixed detector, the transmitted waveform is trained using reinforcement learning based on feedback from the receiver. No prior knowledge is assumed about the target and clutter models. Both transmitter and receiver are implemented as feedforward neural networks. Numerical results show that the proposed end-to-end learning approach is able to obtain a more robust radar performance in clutter and colored noise of arbitrary probability density functions as compared to conventional methods, and to successfully adapt the transmitted waveform to environmental conditions.