Supervised Learning Based Resource Allocation with Network Slicing

With the fast growth of wireless network technologies(e.g., 5G, data center networks) and increasing demand for services with high Quality of Services(QoS), efficient management of network resources becomes more and more important. Network slicing is a effective method for reducing computing time through parallel computing and improve QoS of services. However, determining in which slice a request should be deployed on the premise of ensuring the success rate of transmission is difficult. To this end, we propose a slicing model and a resource allocation scheme in network slices. We make several contributions: i) a slice model that help us reduce the time it takes to calculate routes by parallel computing, and the formulation of resource allocation problem in network slices which aims at maximizing the amount of data transmitted successfully, and ii) a supervised-learning based model to quickly determine in which slices requests should be deployed, that can improve the success rate of transmission. Experimental results show that our proposed approach can achieve a good performance in network slicing environment.