Low-Cost Experimental Platform for AI-Based Beam Management Using WiFi Radios

Valdinei Rodrigues da Conceição, Flavio H F Nunes, João Paulo T Borges, Cleverson Veloso Nahum, Ilan S Correa, Silvia Lins, Aldebaro Klautau

DOI: 10.14209/sbrt.2025.1571157315

Evento: XLIII Simpósio Brasileiro de Telecomunicações e Processamento de Sinais (SBrT2025)

Keywords: Millimeter-wave Beam management Beamforming Robotic

Abstract

Millimeter-wave (mmWave) beam management is critical for 5G/6G networks to address high path loss and ensure reliable directional communication. However, experimental research in this domain faces challenges due to the high cost of specialized hardware, limited accessibility to low-level radio control, and the lack of automated, repeatable testing in dynamic environments. This work introduces a low-cost platform using modified commercial-of-the-shelf (COTS) WiFi radios for AI beam management. Custom firmware enables custom codebooks and real-time metrics like RSSI; a robotic platform automates mobility with synchronized multimodal data collection. Results show it effectively generates rich datasets, validates codebooks, and bridges simulation-reality gaps.

Download