eduzhai > Applied Sciences > Engineering >

Multi-Agent Reinforcement Learning for Adaptive User Association in Dynamic mmWave Networks

  • Save

... pages left unread,continue reading

Document pages: 15 pages

Abstract: Network densification and millimeter-wave technologies are key enablers tofulfill the capacity and data rate requirements of the fifth generation (5G) ofmobile networks. In this context, designing low-complexity policies with localobservations, yet able to adapt the user association with respect to the globalnetwork state and to the network dynamics is a challenge. In fact, theframeworks proposed in literature require continuous access to global networkinformation and to recompute the association when the radio environmentchanges. With the complexity associated to such an approach, these solutionsare not well suited to dense 5G networks. In this paper, we address this issueby designing a scalable and flexible algorithm for user association based onmulti-agent reinforcement learning. In this approach, users act as independentagents that, based on their local observations only, learn to autonomouslycoordinate their actions in order to optimize the network sum-rate. Since thereis no direct information exchange among the agents, we also limit the signalingoverhead. Simulation results show that the proposed algorithm is able to adaptto (fast) changes of radio environment, thus providing large sum-rate gain incomparison to state-of-the-art solutions.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...