AI-Inspired Wireless Communications: Reservoir Computing Meets MIMO-OFDM

# 59






Abstract

Mobile communication has gained lots of interests in recent years as operators across the world try to keep up with the explosive growth in mobile broadband traffic, and to offer innovative new applications and services to continue the momentum of the mobile computing revolution. In this talk, I will provide a brief overview of my research activities related to enabling technologies for 5G and Beyond with a focus on artificial intelligence (AI)-Inspired Wireless Communications. To be specific, I will first discuss top challenges of incorporating AI in wireless communications and introduce the concept of the reservoir computing, a brain-inspired recurrent neural network-based computing methodology, to address some of the key challenges. As a critical task for receive processing, we investigate the application of reservoir computing in the classification task — symbol detection task for MIMO-OFDM systems. Unlike most existing neural network-based symbol detection strategies which require large training sets/overheads, our goal is to design reservoir computing-based symbol detector under limited training sets so that the size of the utilized training set is compatible with those adapted in modern wireless standards (e.g., 3GPP LTE/LTE-Advanced and Wi-Fi). Compared with the benchmark methods (linear minimum mean square error detector and sphere decoder), numerical evaluations suggest that reservoir computing-based MIMO-OFDM symbol detector can improve the symbol detection performance as well as effectively mitigate model mismatch effects using very limited training symbols. Furthermore, we developed a real-time online learning version of the reservoir computing-based MIMO-OFDM symbol detector and implemented the strategy on the software defined radio (SDR) platform for Wi-Fi systems. Keep the Wi-Fi transmitter untouched, the new plug-and-play reservoir computing-based Wi-Fi receiver is comp ared with traditional Wi-Fi receivers using over-the-air experiments. The performance suggests that the introduced symbol detector outperforms conventional Wi-Fi symbol detection methods in various environments indicating the significance and the relevance of the work.

Lingjia Liu, Virginia Tech, USA

Lingjia Liu is a Professor in the Bradley Department of Electrical and Computer Engineering (ECE) at Virginia Tech. He is also serving as the Associate Director in Wireless@Virginia Tech. Prior to joining Virginia Tech, he was an Associate Professor in the EECS Department at the University of Kansas (KU). Before that, he spent 4+ years with Mitsubishi Electric Research Lab (MERL) and Samsung Research America working as a standard delegate in the 3GPP LTE/LTE-Advanced standard. He was a technical leader within Samsung on downlink MIMO, coordinated multipoint (CoMP) transmission/ reception, device-to-device (D2D) commun., and heterogeneous network (HetNet). Lingjia Liu is a senior member of IEEE. He was an Editor for the IEEE Trans. Wireless Commun. and IEEE Trans. Commun. He is currently serving as an Associate Editor for IEEE Trans. Neural Netw. & Learning Syst. He has been serving as the Technical Program Committee Chair of 7 consecutive IEEE GLOBECOM Workshops on Emerging Technologies for 5G (’12-’18). From March 2017 to March 2019, he was elected as the Vice-Chair, Americas of the IEEE Technical Committee on Green Communications & Computing (TCGCC). Currently, he is serving on the Executive Committee of National Spectrum Consortium (NSC) and is an elected member of the IEEE Signal Processing Society SPCOM Technical Committee. Lingjia Liu has 200+ publications including 3 book chapters, 90+ journal articles (most of them are IEEE journals such as TWC, TIT, TCOM, etc), 5 editorials, 100+ conference papers (most of them are IEEE flagship conferences such as GLOBECOM, ICC, GlobalSIP, and ISIT), and 20+ granted U.S. patents. His research received many recognitions in the field including 8 Best Paper Awards. Besides academic research, Lingjia Liu also has numerous technical contributions to the 4G standards including both 3GPP LTE-Advanced and IEEE 802.16m. He has 20+ granted U.S. patents with 10+ patents listed as essential intellectual property rights (IPRs) in 4G standards. Lingjia Liu received the Individual Gold Medal from Samsung and was elected as the 2011 New Faces of Engineering by the National Engineers Week Foundation. His research has been funded by National Science Foundation (NSF), National Spectrum Consortium (NSC), Air Force Office of Scientific Research (AFOSR), Air Force Research Lab. (AFRL), Defense Advanced Research Projects Agency (DARPA), and industry.