11 September 2017 in IEEE Transactions on Vehicular Technology
In this paper, the throughput maximization of millimeter-wave (mm-Wave) ultra-dense networks (UDN) using dynamic spectrum sharing (DSS) is considered. Most of the existing works only allow temporal-domain access and admit at most one user at each time slot, resulting in significant under-utilization of spectrum resource, which will be less attractive to mm-wave UDN applications. A generalized temporal-spatial sharing scheme is proposed in this paper for UDN by exploiting the location information of incumbent devices, where multiple users are allowed to access each channel simultaneously via spatial separations. For distributed applications, the global information exchange among secondary users (SU) tends to be impractical, given the unaffordable signaling overhead and latency. Thus, a non-cooperative game with fine-grained two-dimensional reuse is formulated, which leads to a more efficient access strategy. It is then proved to be an ordinary potential game (OPG), which guarantees the existence of the strategy Nash equilibrium (NE). Finally, an improved decentralized reinforcement learning algorithm is designed, with which SUs can learn from wireless environments and adapt towards to a NE point, relying on the individual observation and the historical action-reward (rather than the global information exchanging). The convergence efficiency of the new scheme is also rigorously proved. Numerical simulations are provided to validate the performances of the proposed schemes.
31 March 2022 in IEEE Transactions on Vehicular Technology
Machine learning has provided great potential in intelligent signal processing, e.g., channel estimation and signal detection. But still, it is difficult to deploy deep neural network (DNN) on low-cost and low-power edge devices, e.g., in both training and inference phases. For one thing, the training of massive network parameters incurs a huge computational complexity and rarely becomes feasible; and for another, even a trained DNN involves still high computations and demands considerable storage size. In this work, we present a tiny machine learning (Tiny-ML) approach for hardware-efficient channel estimation and signal detection. The key innovation of our Tiny-ML is that we replace each large dense layer of DNN with three small cascading sub-layers; and therefore, the computation/storage of a large matrix is replaced with that of small ones. To enable the lightweight training of our Tiny-ML, a novel rank-restricted back-propagation algorithm is further designed. Numerical simulations demonstrate the advantages of our new method. Without sacrificing the estimation and detection accuracy, our Tiny-ML attains an acceleration of $\rm {\mathbf {2.5\sim\!3\times }}$ in model training and $\rm {\mathbf {\sim\!4.5\times }}$ in model inference, as well as a substantial storage reduction of $\rm {\mathbf {\sim\!4.5\times }}$ , when compared to classical fully-connected (FC) DNN. As such, our method paves the way for deploying tiny AI onto low-cost and low-power hardware devices, thereby exploiting the full potential of intelligent signal processing.
25 March 2021 in Nature Computational Science
20 June 2022 in IEEE Transactions on Vehicular Technology
Roadside units (RSUs) can help vehicles sense the traffic environment, so as to improve traffic safety. Since the sensing capability of single RSU is limited, we propose a multiple RSUs cooperative radar sensing network (RSU-CRSN) with signal-level fusion technique. Spatial registration is an essential prerequisite and foundation for RSU-CRSN with signal-level fusion. In this paper, we present an adjustable beam enabled spatial registration algorithm (AB-SRA) that makes the sensing area of each RSU coincide by adjusting the sensing beam width of RSU. To adjust the width of sensing beam flexibly, a beamwidth adjustable beamforming algorithm (BABA) is proposed in this paper. Simulation results show that the performance of AB-SRA is close to perfect spatial registration.
10 April 2023 in Nature Machine Intelligence
Evolutionary computation, for example, particle swarm optimization, has impressive achievements in solving complex problems in science and industry; however, an important open problem in evolutionary computation is that there is no theoretical guarantee of reaching the global optimum and general reliability; this is due to the lack of a unified representation of diverse problem structures and a generic mechanism by which to avoid local optima. This unresolved challenge impairs trust in the applicability of evolutionary computation to a variety of problems. Here we report an evolutionary computation framework aided by machine learning, named EVOLER, which enables the theoretically guaranteed global optimization of a range of complex non-convex problems. This is achieved by: (1) learning a low-rank representation of a problem with limited samples, which helps to identify an attention subspace; and (2) exploring this small attention subspace via the evolutionary computation method, which helps to reliably avoid local optima. As validated on 20 challenging benchmarks, this method finds the global optimum with a probability approaching 1. We use EVOLER to tackle two important problems: power grid dispatch and the inverse design of nanophotonics devices. The method consistently reached optimal results that were challenging to achieve with previous state-of-the-art methods. EVOLER takes a leap forwards in globally guaranteed evolutionary computation, overcoming the uncertainty of data-driven black-box methods, and offering broad prospects for tackling complex real-world problems.
20 October 2022 in IEEE Transactions on Vehicular Technology
Millimeter-wave (mmWave) radar is critical to the emerging automatous driving. As one important step to estimate the range / velocity of unknown targets, constant false-alarm rate (CFAR) techniques should be firstly applied. Most CFAR methods focus on the point -based target model (i.e. each target is represented as one point), which may be inadequate for the highly accurate detection scenarios. Owing to the largely improved temporal/spatial resolutions of mmWave radars, each target is now dispersed to many reflection points, covered by a certain area on Range Doppler Map (RDM). In this work, we fully utilize such new information provided by mmWave radars, and develop an area -based CFAR framework by fully exploiting the potential diversity gain, with which the detection signal-to-noise ratio (SNR) is substantially improved. Theoretical analysis suggests the achieved SNR gain of our method over traditional algorithms grows as the area size $S$ of each target on RDM, i.e. $ {\mathcal {O}}(S)$ . As demonstrated by numerical simulations and real experiments, our method dramatically improves the detection probability of both single-input and single-output (SISO) and multiple-input multiple-output (MIMO) radar systems, also greatly enriching the output point-cloud information for other sophisticated inference tasks. Our method has great potentials in the emerging automotive mmWave radars for highly accurate targets detection.