Developing resource allocation algorithm in device-to-device caching

Evolving technology has triggered the widespread use of online devices and results in an excessive increase in mobile data demand. While meeting this demand service providers must also fulfill users' essential Quality of Service (QoS) requirements such as always-on connectivity, high data rate, and low latency. This exigency conduces to the born of new technologies in the next generation communication systems particularly for 5G and beyond. Device-to-device (D2D) communication is a recent and promising technology that allows edge caching and content delivery between nearby devices. D2D communication exploits the cache capacity of mobile devices and promises content delivery with very low latency and without occupying the cellular network capacity. In this dissertation study, edge caching, content delivery, and device discovery problems in D2D enabled wireless networks are investigated comprehensively in three parts. First, we have derived and solved a Mixed Integer Programming (MIP) model that aims to minimize the Base Station's (BS) total cost with the contribution of data offloading via D2D links. Then we have adopted a distributively implementable Stackelberg Leader-Follower Game algorithm to the corresponding problem and provided the sub-optimal solution. Secondly, the problem of optimal cache placement in a D2D enabled cellular network is considered. The goal of cache placement is to maximize the offloaded traffic. For this purpose, we have derived and solved Mixed Integer Linear Programming (MILP) model with an objective function of maximizing data offloading and burden the load of BS. Optimization-based approaches are not practical in D2D since they provide centralized solutions and bring computational overhead along especially for dense networks. Accordingly, we have proposed two distributively implementable heuristic algorithms. One of them clusters helpers and provides content diversity and cooperation between cache enabled Helper Users (HU). The other algorithm decides the contents to be cached in each cluster, based on the content popularity and intra cluster hierarchy. Lastly, as a preliminary requirement of D2D link establishment, users must be aware of each other and cached contents beforehand. Based on this, we have proposed Discovery and Advertisement-based D2D communication procedures as complementary to the proposed algorithms and applied them to an LTE-based simulation setup. Performance evaluation of all approaches is done according to several varying parameters such as the number of users, the number of contents, content popularity. We also considered interference, cache capacity, and user mobility. Results show that our proposed algorithms provide close to optimal results and outperform basic benchmark schemes even for compelling conditions.