Computer Vision Control for Phased Array Beam Steering

Abstract

This work proves a concept for a wireless access point that uses image identification and tracking algorithms to automate the electronic control of a phased antenna array. Phased arrays change the direction of their radiation electronically by adjusting the phase of the signal applied to the individual antenna elements of the array. This ability can improve a user’s connectivity to a wireless network by directing radiation from an access point to a user, provided that the user’s location is known. Open source image processing and machine learning libraries provided a basis for developing a Python program that determines the position of a target using a single camera. This program uses the position information acquired from the camera to calculate the phases required to steer the radiation of the array to the target. The Python program sends the required phases to another piece of software that controls the phases of the phased array. This software adjusts the phases of the antenna elements and steers the main beam. Experiments were conducted to evaluate the identification, tracking, and control capabilities of the system. Finally, a full system demonstration was performed to benchmark the wireless performance, study the trade-offs in performance for complexity, and compare the connectivity to the current standard in multi-antenna access points.