RRH-Sector selection and load balancing based on MDP and dynamic RRH-Sector-BBU mapping in C-RAN

Abstract

The ever-rising level of traffic in wireless networks is forcing operators to improve significantly their network infrastructure. However, the expected expenditure for building, operating and updating the Radio Access Network (RAN) imposes a challenge since the Average Revenue per User (ARPU) is almost constant or declining slowly. Consequently, mobile operators need to search for solutions that can provide a balance between reaching a satisfactory level of service to customers and costs to optimize the profit and growth. Cloud radio access network (C-RAN) is a novel architecture that supports the tremendous increase in mobile network traffic. Despite the gains that C-RAN offers, a time-varying traffic environment can cause load imbalances, resulting in inefficient resource utilization. Consequently, the network performance can degrade in terms of the blocked users, the number of unnecessary handovers, and the power consumption. This thesis presents an RRH-Sector pair selection for new connections and network load-balancing framework that optimizes the Quality of Service (QoS) and operator reward in C-RAN. In the first part of the framework, we propose a novel RRH-Sector selection algorithm that selects the best RRH-Sector pair for each new connection demand by considering the user and operator objectives. This decision is based on an algorithm derived from the Markov decision process (MDP) with the objective of maximizing the integrated operator-user utility. In the second part of the framework, the loadbalancing problem is addressed via optimization of the RRH-Sector-BBU dynamic mapping formulated as a linear integer-based constrained optimization problem. We compare solutions for this problem obtained by several evolutionary algorithms such as: Bee colony (BCO), Cuckoo search (CUCO), Genetic algorithms (GA), and Particle swarm (PSO). Finally, we evaluate the proposed solutions using extensive simulations. The results show that the proposed RRH-Sector selection scheme provides significant gains in terms of the operator reward and the connection blocking probability when compared to the received signal strength (RSS) method. Furthermore, the evolutionary algorithms are compared with the exhaustive search method that gives the RRH-Sector-BBU optimal mapping. The results show that, in most of the considered scenarios, the proposed algorithms reach the optimal solutions in terms of the number of blocked users, number of handovers, and BBU power consumption. Therefore, the proposed framework enhances the QoS and optimizes the network performance that balances the load across the network.

Date	13 Dec 2021
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Zbigniew Dziong (Supervisor)