Adaptive learning based directional medium access control (MAC) protocol for MillimeterWave communications

Abstract

"The technological advancement demands for more bandwidth and high data rates. These requirement can be fulfilled using millimeter wave band (mmWave) 57-64 GHz which has 7 GHz bandwidth. This ample amount of bandwidth is unlicensed and underexploited whereas traditional wireless network frequency 2.4, 3.6 and 5 GHz band became congested and over used. Recent success in making low cost transceiver at this frequency made communication at 60 GHz commercially attractive. Due to path loss and oxygen absorption at this frequency use of directional antenna is both necessary and practical. Directional antenna reduce interference and increase the spatial reuse. The use of directional antennas possess challenge like making carrier sensing infeasible and deafness in network. In this thesis, our focus is on the design of efficient medium access control (MAC) Protocol for mmWave that overcome the aforementioned challenges posed by directional antenna. We proposed an adaptive learning based directional MAC (ALDMAC) which works on the machine learning algorithm known as Reinforcement Learning (RL) This is a fully distributed and implicit coordination protocol. Specifically, We have conducted performance comparison of AL-DMAC with memory guided directional MAC (MDMAC) and directional slotted ALOHA (DSA). Our simulation results on wireless simulator in MATLAB shows that AL-DMAC has significant increment in throughput compared with MDMAC and DSA. Furthermore, we compared Jain’s Fairness Index of AL-DMAC withMDMACand DSA. The results obtained are promising and shows that AL-DMAC can improve the performance of mmWave wireless communication."