Cognitive algorithms for LTE cloud-RAN

Abstract

In order to maximize the overall system throughput and cope with increasing user density and traffic distribution as also limited spectrum availability in dense urban scenario, next generation small cell networks need to be equipped with dynamic spectrum management schemes through centralized or distributed co-ordination approach. In this paper, we demonstrate using the theory of fluid queues that for a two transmitter case, centralized scheduling of component carrier (CC) of multiple Small Cells (SeNB) of a cellular operator can provide higher achievable throughput as compared to an uncoordinated approach. To quantify the gains in throughput for a realistic cellular setting, we evaluate a greedy CC scheduler on a LTE System Level Simulator and show that up to 40% throughput gains as also energy efficiency gains are possible with joint scheduling. However, these gains are at the expense of the central controller having complete knowledge of the measurement reports of all the cells of each User Equipment (UE). We exploit the correlated nature of the interference in a dense small cell scenario and propose cognitive measurement algorithms considering realistic cellular settings. We show that using our proposed cognitive measurement algorithm, we can decrease the requirement of uplink (UL) overhead from 12.5 UL reports/UE/second to 0.7 UL reports/UE/second while still maintaining a desired gain in throughput over the uncoordinated scheduling scheme.