Regression-based K nearest neighbours for resource allocation in network slicing

Abstract

Network slicing is necessary for modern mobile networks to provide flexible service in areas like smart cities, where there are diverse application requirements as well as growth in demand. In this paper, machine learning K-Nearest Neighbours (KNN) is used to match user distribution scenarios stored within a case library to find out the best boundary between slices without having to perform more computational-expensive approaches. The KNN algorithm is used to identify similar cases and the ratio of qualified users (QUR) who obtained required resources is taken as the test performance indicator.The simulation results show that the proposed architecture is capable of effective slice boundary determination and the resource allocation according to that determination gives good results.