Fast Distributed Beamforming without Receiver Feedback

Kushal Chakrabarti, Amrit S. Bedi, Fikadu T. Dagefu, Jeffrey N. Twigg, Nikhil Chopra

Abstract: Distributed multi-agent beampattern matching inherently enables covert communication by constructively combining the transmitted signals at target receivers and forming nulls towards the adversaries. Most existing beamforming methodologies achieve this by relying on partial or full-state real-time feedback from the receivers. This paper proposes a novel distributed beamforming technique that does not assume any feedback from the receivers or channel parameters, such as multipath fading. The proposed algorithm works in a server-agent architecture of the beamforming agents, eliminating the need for receivers' feedback. Our algorithm is built on the classical gradient-descent (GD) method. However, when the problem is ill-conditioned, GD requires many iterations to converge and is unstable against system noise. We propose an iterative pre-conditioning technique to mitigate the deleterious effects of the data points' conditioning on the convergence rate, facilitating a rapid establishment of communication links with far-field targets. The empirical results demonstrate the proposed beamforming algorithm's favorable convergence rate and robustness against unknown multipath fading in realistic environments.