Performance Analysis of Multi-Armed Bandit Based Resource Allocation Algorithms for D2D Communication Systems
Rahul Shanbhag, Rahul Bajpai, Naveen Gupta, Abhishek Kumar · 2020
Device-to-device (D2D) communications are expected to play an essential role to increase the capacity and throughput of the fifth-generation (5G) wireless communication networks. Efficient resource allocation algorithms are required to be devised in order to accomplish such a task. It is assumed that D2D users are secondary users and it must not provide any interference to the existing cellular users. In this paper, we propose the use of multi-armed bandits (MABs) for developing the resource allocation techniques for dynamic D2D communication systems. Here, we propose four MAB based efficient resource allocation algorithms. Results show that the Thompson Sampling algorithm is a superior algorithm due to its ability to satisfy the regret bounds and provide the optimal reward.