AI-Enabled Network-Functions Virtualization and Software-Defined Architectures for Customized Statistical QoS Over 6G Massive MIMO Mobile Wireless Networks
Xi Zhang, Qixuan Zhu
Abstract
With the rapid deployments of the fifth generation (5G) mobile wireless networks, the shift from the 5G to the sixth generation (6G) mobile wireless networks has attracted tremendous research attention around the world. Featuring with the explosively increasing multimedia-traffics with very diverse services requirements, the 6G mobile wireless networks need to provide the customized services with heterogeneous types of quality of service (QoS) guarantees. However, how to efficiently support these customized services with heterogeneous QoS provisioning for 6G wireless networks has imposed many new challenges not encountered before. To conquer these difficulties, in this article we propose the artificial intelligence (AI)-enabled integration of massive multiple-input-multiple-output (massive-MIMO) techniques with network functions virtualization (NFV) and software-defined network (SDN) architectures to support the customized services over the 6G mobile wireless networks. Specifically, we develop the AI-enabled network architectural schemes which efficiently integrate three 6G-candidate techniques - massive-MIMO, NFV, and SDN - to significantly improve key performances of heterogeneous statistical QoS provisioning in terms of effective capacity. We apply the massive MIMO transmission to substantially improve the channel throughput. Our NFV-based schemes abstract and slice the physical infrastructure and wireless resources in network data plane into several virtualized networks and obtain the optimal service delivery path with the maximum effective capacity among virtualized networks. Also, we develop a set of AI-enabled techniques including multiagent AI-plane architectures, edge-Al frameworks, and federated learning mechanisms for efficiently implementing our developed massive-MIMO-NFV-SDN integrated schemes. Collaborating with our developed platform and techniques, our multi-agent AI-plane based SDN controller coordinates the network nodes and resources allocations for each virtualized network. Our conducted extensive simulations validate and evaluate our developed massive-MIMO-NFV-SDN integrated architectures using AI-techniques, showing that they can efficiently support the customized statistical delay-bounded QoS provisioning over 6G mobile wireless networks.