SoftChain: Dynamic Resource Management and SFC Provisioning for 5G using Machine Learning

Network automation is an area of interest which further incurs in Zero-touch network and Service Management (ZSM). Machine Learning (ML) is acting as a key tool for the realization of such intelligent reformations. Adopting all new technologies inside the network needs enhancement of current orchestration frameworks. Even though existing works are contributing enough to technologically superior countries, this work exclusively aims at developing countries. In this work, a dynamic VNF allocation and embedding problem is addressed on the shared network slices (NSs) using softwarized Service Function Chaining (SFC). The ML techniques are specially designed to make the placement decisions more dynamic and resilient. The proposed VNF-CAR (VNF-Creation, Allocation, and Release) approach is portable among heterogeneous slices and backed by a systematic performance evaluation over a real network topology (INDIA-SDNlib). Accurate VNF selection and embedding are done by supporting OSM MANO decisions (Open Source Management and Orchestration). Through eventful actions, we have shown 20-30% of overall improved VFN selection efficiency against other existing benchmarks.