Energy Detection Adjustment for Fair Coexistence of Wi-Fi and LAA: A Unimodal Bandit Approach

Toward fair coexistence with Wi-Fi, the Listen-Before-Talk (LBT) procedure has been recommended for Licensed-assisted access (LAA) in 3GPP Release 13. The energy detection threshold determines the access opportunity of LAA nodes, which plays a critical role in fairness performance on coexistence network between Wi-Fi and LAA. In this paper, an energy detection dynamic adjustment based on Unimodal Bandits (EDDA-UB) algorithm is proposed to optimize energy detection threshold for LAA eNBs. First, the impact of LAA deployment on Wi-Fi is formulated as the optimal problem with considering the fairness. To solve it, the EDDA-UB algorithm is proposed to learn the optimal energy detection threshold in time varying environment. Simulation results show that the proposed algorithm has a better convergence. Compared with existing constant or adaptive energy detection algorithm, the proposed EDDA-UB algorithm shows the superiority on fair performance and improves the network performance in throughput.