Hybrid Service Function Chains Dynamic Deployment in Multi-Domain Edge Cloud Networks

As the network function virtualization (NFV) tech-nology is introduced, there is an expanding number of Internet-of-Things (IoT) mobile customers leveraging virtual network services in edge cloud networks. NFVs can be connected in a certain order to create a service function chain (SFC), also known as a SF chain. Owing to the large amount of service requests from IoT devices attempting to access their communication networks from anywhere at any time, the requirement for dynamic and adaptive deployment of SFCs to accommodate dynamic changes in service requests is a major challenge today. In this paper, the dynamic deployment of hybrid SFCs in multi-domain edge cloud networks is addressed with the optimisation objective of minimising the total cost. First, we propose an orchestration approach for hybrid SFCs based on splitting and aggregating policies. Therefore, aiming to balance GCN performance and complexity, we introduce an efficient hybrid SFC dynamic de-ployment algorithm, GCN-P-DRL. With extensive experimental evaluation, Our proposed algorithm attains the best performance in most metrics, with a 15.91% improvement in acceptance ratio and a 14.96% reduction in average delay when comparing to the available most advanced solutions.