Smart Sensing and Context-Cognitive Networking In and Beyond mmWave Band: Efficiency, Reliability, and Security

Abstract

The demand for mobile data rates has grown tremendously in recent years, but the current low-frequency spectrum is insufficient to support the rapidly growing demand for data rates. Therefore, the upcoming 5G and future 6G technologies move data transmissions into the unused higher frequency bands for more bandwidth. However, moving the network to a higher frequency band also brings additional difficulties, such as reduced network coverage and reliance on the line-of-sight (LoS) path. Fortunately, today’s smart wireless systems have already transformed from purely communication networks into integrated systems that combine sensing, computing, and communication. This integration facilitates the aggregation of various data streams to form an intelligent and context-aware system that could overcome the limitations presented by high-frequency radio signals.

This dissertation focuses on improving the efficiency, reliability, and security of networks that work in the mmWave band and beyond, with a special perspective of using smart sensing to obtain environmental information and building cognitive networks to obtain context knowledge. In the first work, we explore the environment perception capability of the commercial off-the-shelf (COTS) mmWave device to build a reflector map of the environment. The map is used to find the backup Non-line-of-sight (NLoS) path directions to sustain the communication when the LoS path is blocked. In the second work, we optimize the Reconfigurable Intelligent Surface (RIS)-aided mmWave network topology by considering the directional communication property. A novel (k,a)-Coverage model is proposed to fully characterize the impact of the path difference on the availability of the path. Two deployment schemes are proposed to address the problem of using the least number of RIS to achieve (k,a)-coverage. In the third work, to ensure the data correctness of light detection and ranging (LiDAR) sensors in autonomous vehicles, we propose a Doppler frequency shift-based physical layer spoofing detection method. A statistical spoofing detection framework is also proposed to jointly consider the impact of short-term uncertainty in vehicle velocity.