AI-based Joint Sensing, Communications, and Control for 6G Wireless Networks

点击翻译按钮获取中文摘要

The future 6G wireless networks aim to be a platform for "connected intelligence" and serve as a bridge between the physical world and the cyber world. 6G networks will utilize the sub-6 GHz frequency band as well as the tetrahertz (THz) frequency band. 6G networks can provide global coverage by integrating terrestrial and non-terrestrial communications systems. 6G networks will connect a massive number of heterogenous wireless/mobile devices with sensing capabilities. 6G networks will have the capability to support joint sensing and communications by using the millimeter wave and THz frequency bands and utilizing the base stations as wideband networked sensors. 6G networks will support applications such as augmented reality (AR), autonomous driving, vehicle to everything (V2X), and Industry 5.0. Radio resources in the radio access network and network resources in the core network will be controlled by artificial intelligence (AI) or machine learning algorithms. The 6G network architecture will be distributed with edge computing capabilities for local data collection, training, and inference in order to support new business models and AI-based services. The key performance indicators (KPIs) for 6G networks include the following: peak data rate of 1 Tb/s, user-experienced data rate of 10 - 100 Gb/s, air interface latency of 0.1 ms, localization accuracy of 1 cm for indoor and 50 cm for outdoor environment, and a reliability of 99.99999%. In order to support these applications with different service requirements, fundamental changes to the network architecture and protocol stack are needed. In this proposed research program, the long-term objectives are to enable the emerging 6G networks (a) to support different use cases with diverse KPIs, and (b) to provide native AI and sensing services in a scalable, robust, and spectral/energy-efficient manner. In the short-term, we will focus on three inter-related building blocks: (1) to develop AI-enabled radio access network and core network architecture for new 6G applications including sensing, distributed machine learning, and AI as a service; (2) to develop AI-based air interface, multiple access and scheduling algorithms to support different use cases, terrestrial and non-terrestrial network access, and massive number of heterogeneous devices; and (3) to develop deep learning based integrated sensing, communications, localization, and imaging algorithms. We anticipate that our research will discover practical novel solutions and applications in the information and communications technology (ICT) sector. They are expected to have a lasting impact in contributing towards the long-term competitiveness of Canada's technological innovation.