A Framework for Computer Vision Assisted Beamforming in Aperiodic Phased Arrays

Abstract

Mobile networks, unmanned air vehicles (UAVs), and other dynamic wireless systems are prevalent in several widespread military and commercial applications due dynamic ability to adapt to an environment and commonly implemented autonomous control. Their mobility and dynamic reconfiguration create randomly dispersed networks that inherently give rise to significant electromagnetic challenges in collaborative applications such as beamforming. Conventional radio and communication systems overcome these challenges through designs that are highly structured with respect to frequency and are static in nature.

Two inherent problems prevent collaborative electromagnetic capabilities in these disparate random geometrical systems: cognizance of node positioning and local synchronization of oscillators, phase, and information. This work proposes the combined use of image processing techniques and infrared depth-of-field sensing to detect and track node position in a phased array control framework for morphing clusters of randomly distributed antennas. This framework is presented and designed to uniquely identify array elements (or platforms) and track the motion-dynamic spatial distribution to provide feedback and control information for phase shifting and beamforming. A primary metric of this work is to examine the core performance of the phased array control system with respect to beamforming accuracy. This begins with the use of image recognition algorithms in a reconnaissance phase to establish element identities and determine their locations in the field of view. This process informs the depth-of-field sensor to prompt evaluation of the spatial distribution of elements and enable element location tracking through time. This information is communicated to a distributed array controller that identifies the characteristic function of the array (triangular, spheroidal, etc.), and calculates phases for the elements to achieve the desired beam steering operation. The framework also includes a mobile device (smartphone, tablet, etc.) as a user interface which can be used to control the phased array and link geolocation information for autonomous tracking modes. A framework operating at 2.48 GHz has been developed using low-cost off-the-shelf components, as well as custom-designed element platforms so the performance of the system can be observed experimentally. Results for element identification and spatial distribution are included to benchmark the accuracy of the aforementioned system. Next, a series of experiments demonstrates the operation of the proposed system through radiation patterns that incorporate beam steering and other complex control mechanisms. Analysis of the patterns shows from various geometrical topologies is presented to demonstrate the capability of the system to analyze morphing swarms and clusters. Finally, a conclusion presents findings from some noticeable differences in simulated and measured results.