An Evolutionary Approach for Multiple Flying Base Stations Deployment in NOMA-Based Networks

This paper addresses the problem of deploying a Non-Terrestrial Network (NTN) using multiple Flying Base Stations (FBSs) to serve multiple Stationary Ground Users (SGUs). Our objectives are to minimize the number of active FBSs, reduce overlapping coverage areas, and increase data rates for SGUs. We formulate this as a multi-objective, non-convex Non-Linear Mixed Integer Programming (NLMIP) problem, incorporating both deployment and resource allocation practical constraints. To manage the problem's complexity, we decompose it into two sub-problems: FBS deployment and resource allocation optimization. To optimize the 2D deployment of FBSs in horizontal space, we propose a general, low-complexity heuristic method based on a statistical variant of the Self-Organizing Map (SOM) algorithm, named Evolutionary Bayesian SOM (EBSOM). Then, utilizing a Non-Orthogonal Multiple Access (NOMA) communication scheme between FBSs and SGUs, we maximize the downlink sum rate while adhering to constraints on FBS positions, transmission power, and bandwidth availability. In this stage, the deployment of FBSs is further refined to enhance resource allocation by dividing the problem into two non-convex sub-problems, which are solved iteratively using Surrogate Optimization (SO). Numerical simulations demonstrate the effectiveness of our strategy, particularly in reducing the number of active FBSs, minimizing overlapping areas, and improving network sum rates.