Bayesian Optimization Framework for Channel Simulation-Based Base Station Placement and Transmission Power Design

TL;DR

This paper introduces a Bayesian optimization framework for efficiently designing base station placement and transmission power in wireless networks, maximizing throughput with limited simulation runs.

Contribution

It presents a novel adaptive experimental design method leveraging Bayesian optimization for joint BS placement and power control based on channel simulations.

Findings

Achieves 18-22% higher throughput than traditional methods

Reduces the number of costly channel simulations needed

Provides an effective approach for pre-deployment system configuration

Abstract

This study proposes an adaptive experimental design framework for a channel-simulation-based base station (BS) design that supports the joint optimization of transmission power and placement. We consider a system in which multiple transmitters provide wireless services over a shared frequency band. Our objective is to maximize the average throughput within an area of interest. System operators can design the system configurations prior to deployment by iterating them through channel simulations and updating the parameters. However, accurate channel simulations are computationally expensive; therefore, it is preferable to configure the system using a limited number of simulation iterations. We develop a solver for the problem based on Bayesian optimization (BO), a black-box optimization method. The numerical results demonstrate that our proposed framework can achieve 18-22% higher throughput performance than conventional placement and power optimization strategies.