Abstract
Taking a look over current mobile implementations, the overhead introduced by the uplink signaling can be prevented reducing the rate of Channel Quality Information reports sent from the mobile users to the channel-aware schedulers located at base stations. In this context, the main goal of the present work is to handle the problem of scheduling traffic flows in modern wireless downlink systems under partially observable channel feedback. With this objective, we focus on the design of an scheduler aimed at minimizing the mean flow delay. In this way, we model the optimization problem as a Partially Observable Markov Decision Process, which depends on the belief state of each user’s best channel condition. Due to the complexity of the model, we consider the Whittle index-based relaxation, and we derive an index rule which is easily implementable. Whittle indices are obtained offline by means of an algorithm called Adaptive Greedy. The proposed scheduling policy consists in giving priority to the users with the highest Whittle index value. The resulting scheduling solution is then evaluated in both single-class and multi-class scenarios, using real-world channel quality traces in the performed experiments. The obtained results show that the proposed scheduling index rule outperforms the well-known channel-aware disciplines.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have