The spectrum resources licensed to the mobile operators become increasingly scarce because of the explosive growth of the mobile traffic. Device-to-Device (D2D) communication is thus proposed to be deployed in unlicensed frequency bands, i.e. D2D-Unlicensed (D2D-U). The fixed duty cycle method is generally adopted in the coexistence scenario of D2D and WiFi, which may lead to unfair unlicensed spectrum usage since it cannot adapt the data traffic change. Therefore, a Q-learning (QL) based resource allocation algorithm for D2D-U is proposed in this paper. In the algorithm, the considered cellular base station acts as the agent. The actions of agent are defined as the different combinations of the transmission power and the duty cycle of D2D-U users, and the states of agent are defined as the different combinations of the total throughput, fairness and signal-to-noise ratio (SNR) of cellular users. Based on the proposed QL framework, the agent can always learn the optimal power allocation and duty cycle by interacting with the environment, which can maximize the total throughput and fairness while ensuring the satisfactory SNR of cellular users. The simulation results show that the proposed algorithm can obtain the largest throughput and the best fairness while ensuring the satisfactory SNR of LTE-U users among all traditional algorithms.
A Q-learning based Resource Allocation Algorithm for D2D-Unlicensed communications
01.04.2021
1905680 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Meta-Reinforcement Learning Based Resource Allocation for Dynamic V2X Communications
BASE | 2021
|Unlicensed Technology Assessment for Uas Communications
IEEE | 2020
|