Scalable and QoS-Aware Resource Allocation to Heterogeneous Traffic Flows in 5G

Abstract

Networks of new generations are increasingly involved in transporting heterogeneous flows. Indeed, in addition to the usual data and multimedia traffic, the Internet of Things (IoT) smart applications are creating new traffic types and relationships involving billions of active nodes like sensors and actuators. This traffic raises a problem of scale, particularly for resource management and decision-making mechanisms. The present work addresses for the first time the joint problem of mapping heterogeneous flows from multiple users and applications to transport blocks, and then packing these blocks into the rectangular grid of time–frequency resources within a flexible 5G new radio frame. Our solution is based on a quality-of-service-based classification of flows followed by an offline construction of two databases. The first one enumerates all possible configurations of transport blocks and the second enumerates all possible configurations of frames. Thus, the sole online processing that remains to be done is to find the optimal block configurations that satisfy a given request vector. Hence, the resolution of this complex joint mapping and packing problem is reduced to a simple resolution of a linear problem, which consists in finding the best configurations. A thorough numerical study shows that our configuration-based solution can map, within few tens of milliseconds, more than 100 flow connections to transport blocks incurring only 3% of overallocation, and then pack these blocks into the grid leading to an upper bound on the optimality gap as low as 2.8%.