User-centric strategies for resource management in heterogeneous wireless networks with QoS considerations

Abstract

Demand for mobile applications is increasing at an exponential rate which is loading existing wireless networks. The research community is currently actively involved in the design of new technologies that can enable massive device connections with the needed speeds and reliability. To this end, a major opportunity is to design solutions that facilitate the dynamic utilization and seamless operation of heterogeneous networks where devices can utilize multiple wireless interfaces simultaneously and cooperate with other devices in their vicinity. In this thesis, we propose and evaluate novel solutions to address emerging challenges related to the design of next generation heterogeneous wireless networks. Our research work is divided into two key objectives: the first objective aims at designing e↵ective user-centric resource management techniques in cellular/WiFi heterogeneous networks with quality of experience considerations, and the second objective aims at optimizing trac o✏oading in highly dense wireless networks using device-to-device cooperation, local caching, and planned channel allocation. To achieve the first objective, we propose cellular/WiFi resource management strategies for a single-user scenario where a user can take advantage of the coexistence of multiple wireless interfaces to achieve performance gains. We first design a learning-based approach for network selection where a user utilizes one wireless interface at a time to achieve either minimum energy consumption, maximum throughput or energy eciency based on user preferences. We then formulate the static trac splitting problem, where a user utilizes both interfaces simultaneously, as a multi-objective optimization approach that captures the tradeo↵s between throughput maximization on one hand and device battery energy minimization on the other hand. We then extend our work to address real-time trac splitting decisions capturing the tradeo↵ between queue stability, energy consumption, and quality of experience for video streaming applications. The proposed strategies are evaluated using simulations and experimental test bed measurements under realistic operational conditions. In the second thesis objective, we propose multi-user trac o✏oading strategies for dense wireless networks where a very high number of users in a given geographical area request simultaneously access to given data services, e.g., in a sports stadium or exhibition center. We formulate the problem as an optimization that aims at maximizing the number of served users while maintaining target quality of service using device-to-device cooperation, in-device caching, and intelligent channel allocation. Due to the complexity of the problem, we design sub-optimal hierarchical tree-based algorithms for real-time operation taking into account realistic constraints. We demonstrate their e↵ectiveness by presenting performance results and analysis for a wide range of network scenarios.