Concept Design of Medium Access Control for Spectrum Access Systems in 3.5 GHz

The issue of finding a three-tiered spectrum access scheme in the so-called innovation band from 3550 to 3700 MHz for incumbent applications and protection against other priority access licenses (PAL) and general authorized access (GAA) systems has been addressed by the federal communications commission (FCC). The PAL and GAA systems are citizens broadband radio service devices (CBSDs) connected to a spectrum access system (SAS) which controls the spectrum access. The challenge lies in creating an approach, where the CBSDs are able to fulfil interference mitigation (IM) among each other. In this paper, we present the design and results of a hybrid centralized and distributed medium access control (HMAC) scheme for SAS with efficient and self-organized spectrum sharing schemes using machine learning algorithms based on game theory (GT) and reinforcement learning (RL) for GAA CBSDs in an unlicensed spectrum. The system improvement shows how the SAS can assists decision-making algorithms in a minimum intrusive way to fulfill fair interference and channel occupancy compared with state-of-the-art (SoA) schemes, where the SAS uses an interference graph (IG) to solve the channel allocation problem. Moreover, the suggested schemes support the distributed carrier sense multiple access/collision avoidance (CSMA/CA) mechanism to handle adjacent-channel and cochannel interference (ACI, CCI), which opens new perspective for decentralized systems in a SAS context and beyond 5G.