Intelligent Deployment and Orchestration of E2E Slices

Realizing multi-domain end-to-end (E2E) slices in the federated infrastructure network requires efficient slice deployment policy and slice orchestration mechanism. An E2E slice which stretch across the radio access network (RAN), transport, and core network need to be deployed rapidly in different administrative domains and the resources should be allocated to slices for satisfying differentiated service requirements. Hence, this chapter introduces a service-oriented slice deployment policy which takes into consideration the characteristics of service requirements and designs dedicated policies for different types of services. However, in the era of 6G, the increase of service types and the dynamic change of service requirements bring new challenges to slice management and orchestration. To handle the dynamic and complicated slice requests from verticals, incorporating artificial intelligence (AI) into slice management and orchestration has become a tend. In this chapter, an AI-enabled slice orchestration framework is proposed to enable intelligent admission control of various slice requests from multiple tenants and deal with the changes in resource requirements of slices. This framework adopts deep reinforcement learning (DRL) to improve the resource efficiency from a global perspective while perform slice adaption in real-time manner. Furthermore, an advanced DRL-based method is designed in slice reconfiguration in order to realize fast and optimal decision-making in resource adjustment for multiple types of E2E slices.