Radio Resource Management for Rate Maximization with QoS/QoE Provisioning in Wireless Networks

Abstract

Currently, the Fifth Generation (5G) of mobile communications is under intensive discussions in order to setup the new standardization. It is already known that 5G systems must provide support to a large variety of applications besides handling a higher number of devices connected to the network. One of the use cases of the 5G is the Enhanced Mobile Broadband (eMBB), which consists in an improvement of the existing Fourth Generation (4G) broadband service. The eMBB focuses on providing, among other characteristics, high system capacity, high peak data rate and user experienced data rate. One possible strategy to achieve the eMBB goals is to properly use Radio Resource Allocation (RRA) methods to increase the efficiency of the spectrum usage and the Quality of Service (QoS) perceived by the users. Therefore, the work developed in this thesis studies methods of RRA aiming at maximizing the overall system throughput, constrained by guaranteeing a certain satisfaction rate per service in single and multi-service scenarios. The RRA problems addressed in this thesis deal with different service requirements and a trade-off between high spectral efficiency and users satisfaction. This trade-off can be managed by the system, which makes the study performed in this thesis very relevant, mainly to the mobile network operators. The RRA methods are studied in three different contexts. The first problem considers that the Base Station (BS) employs an Equal Power Allocation (EPA) among Resource Blocks (RBs) and only the RB assignment is addressed by the RRA. Besides, the users shall meet their requirements in a single time slot, i.e., on a Transmission Time Interval (TTI) basis. This problem is initially described mathematically and, from the analysis of the optimization problem formulation, a new suboptimal low complexity heuristic is proposed. By means of

computational simulations, it is shown that the proposed algorithm outperforms the state-of-the- art heuristic, achieving near optimal results. Moreover, in contrast to the state-of-the-art literature,

the proposed algorithm is capable of providing near feasible solutions in infeasible instances of the problem. Thereafter, this RRA problem is extended to address the users’ requirements over a given timespan. In this context, the heuristic earlier proposed is extended to schedule the users over time. The proposed heuristic is compared with utility-based benchmark algorithms with similar objectives. Simulation results show that proposed scheduler considerably outperforms the benchmark solutions in terms of both satisfaction and overall system throughput. Lastly, the RRA problem is once again studied on a TTI basis, however allocating both power and RBs. Like the first problem studied in this thesis, the RRA is firstly stated as an optimization problem. Using the same solution framework adopted with the first problem, a new suboptimal heuristic is proposed. Computational simulations show that the proposed heuristic outperforms the state-of-the-art algorithm. Additionally, the proposed heuristic is capable of providing near feasible solutions in infeasible instances of the RRA problem, while the state-of-art literature does not provide a practical solution.