A Bidirectional Alignment Control Approach for Planar LED-based Free-Space Optical Communication Systems

Abstract

Achieving and maintaining line-of-sight (LOS) is essential for free-space optical (FSO) communication systems due to the high directionality of the optical signals. We consider the problem of achieving LOS in a planar setting, where each of the two parties (called agents) makes a move only based on its own light intensity measurement. The problem is formulated as a discrete time dynamical system, where each agent seeks to maximize its own reward function that depends on the states of both agents, who make their moves in parallel. In particular, the reward function corresponds to the light intensity measurements made by each agent. While the two reward functions are non-conflicting (i.e., the optimization of one reward function helps the optimization of the other), the constraints of no information exchange between the agents, no access to states, and parallel (non-sequential) actions pose significant challenges. A novel iterative optimization algorithm meeting all of these constraints is proposed. We show that the proposed algorithm brings the system to a neighborhood of the LOS in a finite number of steps, when the reward functions are of the Gaussian form as supported by the experimental data. The effectiveness of the approach is evaluated in simulation by comparison with extremum-seeking control, where it shows an order of magnitude better performance in terms of the convergence speed.