Scheduling the Data Transmission Interval in IEEE 802.11ad: A Reinforcement Learning Approach

The IEEE 802.11ad Wi-Fi standard enables communications in the unlicensed mm-wave band at 60 GHz. Propagation at such frequencies accounts for increased path loss and sensitivity to blockage when compared to the traditional sub-6-GHz Wi-Fi frequencies. To cope with these phenomena, directional transmissions through beamformed links are needed. Many new features have been introduced with IEEE 802.11ad in order to realize directional communications in this band. One of the most crucial changes compared to legacy Wi-Fi communication paradigms is the introduction of a hybrid Medium Access Control (MAC), which enables contention-free and contention-based channel access. The increased complexity associated with hybrid channel access at the MAC layer must be addressed through the development of a resource scheduling algorithm. This paper proposes two novel resource scheduling mechanisms for IEEE 802.11ad. The first approach serves as a baseline for the development of a more advanced strategy based on Reinforcement Learning (RL). Indeed, the second scheme exploits RL to successfully find the optimal duration of each contention-free access period. Our performance evaluation shows that the policy based on RL provides the same level of expected throughput and delay performance while preserving more transmission time to be devoted to other traffic in order to enhance the network efficiency.