In this report, we refer to our previous work [7], where the problem of cell selection in open-access femtocell networks was formulated as a reinforcement learning (RL) framework in a non-stationary scenario. The user selects, without prior knowledge about the environment, a target cell by exploring past cells' behavior and predicts their potential future states based on Q-learning algorithm. Then, Q-learning was extended by referring to a fuzzy inference system (FIS) to tune learning parameters during the learning process to adapt to environment changes. In [7], the solution aims at minimizing the frequency of handovers without affecting the user experience in terms of user capacity. In this report, we investigate the effect of user-speed's variation on learning-based cell selection for femtocell network. Also, we present a solution to counteract the negative effect of user-speed variation on learning. Simulation results show that ・user-speed variation results in reward degradation during the simulation. ・the presented solution offsets the effect of variable speed by increasing the learning periods.
View full abstract