Cost-Efficient and QoS-Aware User Association and 3D Placement of 6G Aerial Mobile Access Points

TL;DR

This paper presents a cost-efficient, QoS-aware framework for 3D deployment of mobile access points and user association in 6G networks, optimizing coverage, quality, and energy use in dynamic scenarios.

Contribution

It introduces an iterative deployment algorithm and a multi-agent reinforcement learning approach for user association, addressing interference, mobility, and energy efficiency in 6G MAP deployment.

Findings

Optimized MAP deployment improves network coverage and QoS.

Distributed user association maximizes network sum-rate.

Proposed methods adapt efficiently to dynamic network conditions.

Abstract

6G networks require a flexible infrastructure to dynamically provide ubiquitous network coverage. Mobile Access Points (MAP) deployment is a promising solution. In this paper, we formulate the joint 3D MAP deployment and user association problem over a dynamic network under interference and mobility constraints. First, we propose an iterative algorithm to optimize the deployment of MAPs. Our solution efficiently and quickly determines the number, position and configuration of MAPs for highly dynamic scenarios. MAPs provide appropriate Quality of Service (QoS) connectivity to mobile ground user in mmwave or sub-6GHz bands and find their optimal positions in a 3D grid. Each MAP also implies an energy cost (e.g. for travel) to be minimized. Once all MAPs deployed, a deep multiagent reinforcement learning algorithm is proposed to associate multiple users to multiple MAPs under interference constraint. Each user acts as an independent agent that operates in a fully distributed architecture and maximizes the network sum-rate.