Collaborative Research: NeTS: JUNO3: SWIFT: Softwarization of Intelligence for Efficient 6G Mobile Networks

点击翻译按钮获取中文摘要

Future wireless networks are expected to operate in the 60 GHz WiGig frequency band to support demanding applications such as virtual reality and augmented reality. However, the wireless links at the WiGig band suffer from frequent outages. This in turn degrades the perceived service quality for mobile users. To enable stable and high-quality wireless connectivity in the WiGig band, this joint US-Japan project proposes an intelligent layer that learns the conditions of the wireless links and network traffic load and makes informed network management decisions to mitigate network disruption. This layer will support ultra-high data rates with low latency, thus ensuring good application performance and user satisfaction despite the wide dynamic range present in the environment. By enabling stable and high-quality WiGig connections, this project will enable sixth generation (6G) applications such as haptic communication, augmented reality, virtual reality, remote surgery, etc. Hence, this project broadly impacts several aspects of society and enhances the economic competitiveness of the U.S.In the future, ultra-high throughput and ultra-low delay applications are anticipated to constitute 90% of mobile data traffic. As a result, 6G wireless networks will operate in the uncongested high frequency bands, e.g., the 60 GHz WiGig. Due to limited diffraction capabilities, the wireless links at the WiGig band suffer from frequent outages. To maintain stable wireless connectivity with user mobility, this joint US-Japam project proposes SWIFT, SoftWarization of Intelligence for eFficient 6G mobile neTworks. SWIFT extends the software defined networking (SDN) architecture by integrating an artificial intelligence (AI)-based layer in the SDN control plane to enable efficient resource management decisions. The project includes the following research thrusts: (1) Generation, validation, and characterization of wireless channel gain and traffic load in WiGig networks while considering dynamic scenarios; (2) Development of efficient prediction models for wireless channel gain and traffic load in WiGig networks based on deep machine learning techniques; (3) Development of AI-based strategy for channel assignment in WiGig networks based on reinforcement learning tools. The project plans a proof-of-concept implementation and performance validation of the SWIFT framework using a state-of-the-art testbed that mimics indoor mobile WiGig networks.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.