Abstract:Multi-arm bandit (MAB) algorithms have been used to learn optimal beams for millimeter wave communication systems. Here, the complexity of learning the optimal beam linearly scales with the number of beams, leading to high latency when there are a large number of beams. In this work, we propose to integrate radar with communication to enhance the MAB learning performance by searching only those beams where the radar detects a scatterer. Further, we use radar to distinguish the beams that show mobile targets from those which indicate the presence of static clutter, thereby reducing the number of beams to scan. Simulations show that our proposed radar-enhanced MAB reduces the exploration time by searching only the beams with distinct radar mobile targets resulting in improved throughput.
Abstract:Millimeter wave (mmW) codesigned 802.11ad-based joint radar communication (JRC) systems have been identified as a potential solution for realizing high bandwidth connected vehicles for next-generation intelligent transportation systems. The radar functionality within the JRC enables accurate detection and localization of mobile targets, which can significantly speed up the selection of the optimal high-directional narrow beam required for mmW communications between the base station and mobile target. To bring JRC to reality, a radar signal processing (RSP) accelerator, co-located with the wireless communication physical layer (PHY), on edge platforms is desired. In this work, we discuss the three-dimensional digital hardware RSP framework for 802.11ad-based JRC to detect the range, azimuth, and Doppler velocity of multiple targets. We present a novel efficient reconfigurable architecture for RSP on multi-processor system-on-chip (MPSoC) via hardware-software co-design, word-length optimization, and serial-parallel configurations. We demonstrate the functional correctness of the proposed fixed-point architecture and significant savings in resource utilization (~40-70), execution time (1.5x improvement), and power consumption (50%) over floating-point architecture. The acceleration on hardware offers a 120-factor improvement in execution time over the benchmark Quad-core processor. The proposed architecture enables on-the-fly reconfigurability to support different azimuth precision and Doppler velocity resolution, offering a real-time trade-off between functional accuracy and detection time. We demonstrate end-to-end RSP on MPSoC with a user-friendly graphical user interface (GUI).
Abstract:Rapid beam alignment is required to support high gain millimeter wave (mmW) communication links between a base station (BS) and mobile users (MU). The standard IEEE 802.11ad protocol enables beam alignment at the BS and MU through a lengthy beam training procedure accomplished through additional packet overhead. However, this results in reduced latency and throughput. Auxiliary radar functionality embedded within the communication protocol has been proposed in prior literature to enable rapid beam alignment of communication beams without the requirement of channel overheads. In this work, we propose a complete architectural framework of a joint radar-communication wireless transceiver wherein radar based detection of MU is realized to enable subsequent narrow beam communication. We provide a software prototype implementation with transceiver design details, signal models and signal processing algorithms. The prototype is experimentally evaluated with realistic simulations in free space and Rician propagation conditions and demonstrated to accelerate the beam alignment by a factor of four while reducing the overall bit error rate (BER) resulting in significant improvement in throughput with respect to standard 802.11ad. Likewise, the radar performance is found to be comparable to commonly used mmW radars.
Abstract:Narrowband and broadband indoor radar images significantly deteriorate in the presence of target dependent and independent static and dynamic clutter arising from walls. A stacked and sparse denoising autoencoder (StackedSDAE) is proposed for mitigating wall clutter in indoor radar images. The algorithm relies on the availability of clean images and corresponding noisy images during training and requires no additional information regarding the wall characteristics. The algorithm is evaluated on simulated Doppler-time spectrograms and high range resolution profiles generated for diverse radar frequencies and wall characteristics in around-the-corner radar (ACR) scenarios. Additional experiments are performed on range-enhanced frontal images generated from measurements gathered from a wideband RF imaging sensor. The results from the experiments show that the StackedSDAE successfully reconstructs images that closely resemble those that would be obtained in free space conditions. Further, the incorporation of sparsity and depth in the hidden layer representations within the autoencoder makes the algorithm more robust to low signal to noise ratio (SNR) and label mismatch between clean and corrupt data during training than the conventional single layer DAE. For example, the denoised ACR signatures show a structural similarity above 0.75 to clean free space images at SNR of -10dB and label mismatch error of 50%.