15 May 2018 in IEEE Transactions on Wireless Communications
Massive multiuser multiple-input multiple-output (MU-MIMO) systems are expected to be the core technology in fifth-generation wireless systems because they significantly improve spectral efficiency. However, the requirement for a large number of radio frequency (RF) chains results in high hardware costs and power consumption, which obstruct the commercial deployment of massive MIMO systems. A potential solution is to use low-resolution digital-to-analog converters (DAC)/analog-todigital converters for each antenna and RF chain. However, using low-resolution DACs at the transmit side directly limits the degree of freedom of output signals and thus poses a challenge to the precoding design. In this study, we develop efficient and universal algorithms for a downlink massive MU-MIMO system with finitealphabet precodings. Our algorithms are developed based on the alternating direction method of multipliers (ADMM) framework. The original ADMM does not converge in a nonlinear discrete optimization problem. The primary cause of this problem is that the alternating (update) directions in ADMM on one side are biased, and those on the other side are unbiased. By making the two updates consistent in an unbiased manner, we develop two algorithms called iterative discrete estimation (IDE) and IDE2: IDE demonstrates excellent performance and IDE2 possesses a significantly low computational complexity. Compared with state-of-the-art techniques, the proposed precoding algorithms present significant advantages in performance and computational complexity.
1 February 2022 in IEEE Transactions on Biomedical Circuits and Systems
During the global epidemic, non-contact methods for monitoring the vital signs of several people have become particularly important. Advanced signal processing techniques have recently been demonstrated to separate and track the vital signs of multiple people. In this paper, we further develop the multi-person vital signs identification (VSign-ID) system to make non-contact detection available in public places. VSign-ID not only extracts multi-person vital signs but also states from whom these vital signs are collected. We utilize multiple doppler radars to expand the effective range of the measurement area and propose a space and time matching mechanism for vital signs identification. We use a thermal camera to detect the number of people and their movements. VSign-ID efficiently coordinates these two types of sensors (i.e., the doppler radars and the thermal camera) to track and identify the respiration rates and heartbeat rates of multiple people. A series of experiments and simulations are conducted to measure the efficiency of VSign-ID. In the case of five people sitting closely, the estimation errors for respiration and heartbeat rates are −4.85 dB and −2.36 dB lower than the standard resolution of the system, respectively, despite using only two independent radars.
21 September 2020 in IEEE Access
Human pose estimation is an important task for several applications, such as video surveillance systems. However, color (i.e., RGB) images may not always be available under certain conditions, such as privacy issues and lack of illumination. In these scenarios, thermal images are more prominent than color images. We introduce in this study ThermalPose, which is a neural network system that parses thermal images and extracts accurate 2D human poses. ThermalPose uses lightweight neural network models that can be easily matched to the design requirements for Internet-of-Things applications. The performance of ThermalPose in visible scenes only slightly decreases compared with that of the state-of-the-art vision-based pose estimator. Meanwhile, in complex scenes with masking background textures or lack of illumination, ThermalPose does not degenerate in terms of performance, whereas the vision-based system completely fails.
30 October 2018 in IEEE Transactions on Wireless Communications
A receiver architecture with low-resolution analogto- digital converters (ADCs) coupled with large antenna arrays has drawn considerable interest in the millimeter wave (mmWave) system. Although architecture with pure one-bit ADCs has low power cost, such a system presents many challenges for synchronization, channel estimation, and power level estimation. Recent research has been conducted on the so-called mixed one-bit system, in which most antennas are equipped with one-bit ADCs, and a few have high-resolution ADCs. Despite the advantages of this system, studies on how to efficiently use high-resolution outputs to aid the one-bit system are lacking. This research considers the channel estimation problem to fill this gap and proposes a two-step channel estimator by utilizing different features of mixed outputs. The channel gain can be extracted from high-resolution ADCs, and the channel angle can be extracted by the combination of additional one-bit ADCs. The proposed estimator leverages the sparsity feature of the channel in mmWave. In contrast to previous works that used compressive sensing techniques, which confine the estimate to the set of grid angle points and induce estimation bias, the proposed estimator is gridless and treats the angle as a continuous parameter. Simulation results demonstrate that the proposed method yields significantly lower mean square errors than the conventional maximum likelihood estimator. In addition, this study investigates ways to further improve the channel estimate by arranging the locations of high-resolution ADCs. Several useful observations on system design are obtained through our analysis.
21 February 2020 in IEEE Transactions on Signal Processing
In this study, we present a phase retrieval solution that aims to recover signals from noisy phaseless measurements. A recently proposed scheme known as generalized expectation consistent signal recovery (GEC-SR), has shown better accuracy, speed, and robustness than many existing methods. However, sensing high-resolution images with large transform matrices presents a computational burden for GEC-SR, thereby limiting its applications to areas, such as real-time implementation. Moreover, GEC-SR does not support distributed computing, which is an important requirement to modern computing. To address these issues, we propose a novel decentralized algorithm called “deGEC-SR” by leveraging the core framework of GEC-SR. deGEC-SR exhibits excellent performance similar to GEC-SR but runs tens to hundreds of times faster than GEC-SR. We derive the theoretical state evolution for deGEC-SR and demonstrate its accuracy using numerical results. Analysis allows quick generation of performance predictions and enriches our understanding on the proposed algorithm.