Peter LoPresti

This work presents a reinforcement learning (RL) approach for optimizing beam intensity in free-space optical communication (FSOC) links under thermal turbulence. A Twin Delayed Deep Deterministic Policy Gradient (TD3) agent controls Zernike phase parameters on a high-resolution spatial light modulator (SLM) to maximize received power. Experiments at 72°F, 240°F, and 280°F demonstrate consistent convergence and robust performance across varying turbulence levels, with convergence achieved in under 2.7 minutes and average power gains of 2–2.7 dB relative to a flat phase. The method is designed for next-generation low Earth orbit (LEO) optical links, where it can be applied to compact ground station terminals or feeder links to provide real-time adaptive phase correction and maintain stable communication under dynamic channel conditions.

Atmospheric turbulence severely limits the stability and reliability of free-space optical (FSO) uplinks by inducing wavefront distortions and random intensity fluctuations. This study investigates the use of reinforcement learning (RL) with beacon-based feedback for adaptive beam shaping in a spatial light modulator (SLM)-controlled FSO link. The RL agent dynamically adjusts phase patterns to maximize received signal strength, while the beacon channel provides turbulence estimates that guide the optimization process. Experiments under low, moderate, and high turbulence levels demonstrate that incorporating beacon feedback can enhance link stability in severe conditions, reducing signal variability and suppressing extreme fluctuations. In low-turbulence scenarios, the performance is comparable to non-feedback operation, whereas under high turbulence, beacon-assisted control consistently achieves lower coefficients of variation and improved bit error rate (BER) performance. Under high turbulence replay experiments—where the best-performing RL-learned phase patterns are reapplied without learning—further show that configurations trained with feedback retain robustness, even without real-time turbulence measurements under high turbulence. These results highlight the potential of integrating contextual feedback with RL to achieve turbulence-resilient and stable optical uplinks in dynamic atmospheric environments.

Fiber-bundle-based free-space optical receivers have potential to increase the receiver field of view and mitigate turbulence effects on collected power. Combining optical signals from individual fibers into a single data signal remains a challenge. We present a homogenizer-based method for optically combining individual signals and compare this to a lens-based method. A simulation assessed signal combining properties of the homogenizer system. Experiments compared the two methods on the field of view (FOV) and bit-error-rate (BER) versus the signal to noise ratio (SNR) for a range of turbulence levels. The homogenizer-based method increases the FOV by at least three degrees. The BER increased more slowly with increasing turbulence for the homogenizer-based method than for the lens-based design and was lower for all turbulence levels studied. Thus, the homogenizer-based method provides a potentially viable solution for addressing the signal combination problem in fiber-bundle-based receivers.

Radiofrequency (RF) communications offer reliable but low data rates and energy-inefficient satellite links, while free-space optical (FSO) promises high bandwidth but struggles with disturbances imposed by atmospheric effects. A hybrid RF/FSO architecture aims to achieve optimal reliability along with high data rates for space communications. Accurate prediction of dynamic ground-to-satellite FSO link availability is critical for routing decisions in low-earth orbit constellations. In this paper, we propose a system leveraging ubiquitous RF links to proactively forecast FSO link degradation prior to signal drops below threshold levels. This enables pre-calculation of rerouting to maximally maintain high data rate FSO links throughout the duration of weather effects. We implement a supervised learning model to anticipate FSO attenuation based on the analysis of RF patterns. Through the simulation of a dense lower earth orbit (LEO) satellite constellation, we demonstrate the efficacy of our approach in a simulated satellite network, highlighting the balance between predictive accuracy and prediction duration. An emulated cloud attenuation model is proposed to provide insight into the temporal profiles of RF signals and their correlation to FSO channel dynamics. Our investigation sheds light on the trade-offs between prediction horizon and accuracy arising from RF beacon numbers and proximity.

Multi-user free-space optical communication (FSOC) is beginning to draw a significant attention for its ability to support increased system network capacity while using single receiving photodiode and satisfying size, weight, and power (SWaP) constraints imposed by space- and aerial-based mobile communication. Despite these advantages, support of multi-user capabilities cause increased system complexity due to accommodating heterogenous users communications with varying transmission and data rate requirements. Machine learning (ML) has recently been considered as a promising approach for introducing cognition into the network to mitigate some of the complexity. A cognitive method based on unsupervised ML was derived for estimating the number of users communicating and sharing time and bandwidth resources with a single-node receiver. A weighted clustering approach was introduced and experimentally validated when users received with similar amplitude information that results in underestimation. Obtained results confirmed that the proposed methodology was able to accurately differentiate the number of simultaneously transmitting users with accuracy greater than 92%— even in the presence of moderate atmospheric turbulence. An experimental analysis was conducted to determine data size and receiver sampling rate requirements for accurate estimation. Furthermore, an empirical model was derived to evaluate the effect of preamble signal length given a particular sampling rate on the accuracy of the estimation. The model was validated for up-to four users.

Multi-point free-space optical communication (FSOC) has recently received extensive interest as a valuable solution for providing increased capacity in the upcoming communication networks. Diverse optical multiple access control (O-MAC) techniques have proved to successfully support multi-user FSOC. However, to adaptively and automatically select the most fitting O-MAC technique or combination thereof for signal demodulation, the number of transmitting users should be known a priori at the receiver side. In this letter, a technique based on histogram peak detection and unsupervised machine learning (e.g., k-mean, k-medoid, hierarchical, and fuzzy clustering), is proposed and experimentally demonstrated for number of users detection. The proposed methodology relies on amplitude information of received signals. Results show that when multiple users— signals are received with equal amplitude, under-estimation of user number was observed; hence, a modified approach—a weighted clustering—was employed and experimentally validated. The work reported herein demonstrated the ability of the proposed methodology to simultaneously and accurately detect multiple transmitting users in real time and under severe atmospheric turbulence.

A fiber-bundle approach for beam steering at the transmitter was recently proposed and investigated that allowed tracking of the receiver without the use of mechanical devices for mobile free-spaced optical (FSO) platforms experiencing misalignment due to movement and atmospheric turbulence. This paper investigates different transmitter configurations along with a general pointing, acquisition and tracking (PAT) algorithm on key parameters including power availability at the receiver, switching time, link recapturing time, and coverage area at the receiver plane. The alignment control algorithm with different ways to track receiver is also evaluated on these key parameters. The algorithm performance versus transmitted power, switching time and the threshold power for beam steering was examined. The investigation finds that a combination of switching combined with splitting of the light between fibers is a suitable solution to provide the best trade-off between link up time and the coverage area at the receiver, and thus is a good general purpose design choice, and the optical tracking system with target location information demands fewer efforts to recover and maintain the dynamic link than the one without the awareness of the receiver location information.