Loading [a11y]/accessibility-menu.js
Multi-UAV mmWave Beam Tracking using Q-Learning and Interference Mitigation | IEEE Conference Publication | IEEE Xplore

Multi-UAV mmWave Beam Tracking using Q-Learning and Interference Mitigation


Abstract:

Applications of UAVs have attracted attention recently. Some operations are subject to a low-latency constraint due to high-speed UAVs in 3D space. As a result, low-laten...Show More

Abstract:

Applications of UAVs have attracted attention recently. Some operations are subject to a low-latency constraint due to high-speed UAVs in 3D space. As a result, low-latency and directional mmWave communications can be taken into account in UAV communications. To overcome mmWave hardware limitations and severe environmental conditions, UAVs can exploit hybrid analog-digital beamforming to improve data rates. For the purpose of being flexible in mission-oriented deployment, the movement of UAVs in 3D space is common but makes beamforming more challenging. The beamforming problem can be solved into two steps: predicting analog beams and then optimizing the corresponding digital weights in terms of SINR maximization. Borrowing the idea of machine learning (especially Q-learning) that uses experience to solve the prediction problem, we implement Q-learning-based beam prediction using coupling coefficients. Compared with other beam tracking methods given knowledge of channel or SINR, using coupling coefficients as observations is a more feasible and efficient solution to rapid beam tracking. The coupling coefficients can be further used to approximate SINR measurements given the selected analog beams. Using the proposed relationship between the SINR measurements and digital weights, we can easily obtain the optimal digital weights.
Date of Conference: 07-11 June 2020
Date Added to IEEE Xplore: 21 July 2020
ISBN Information:

ISSN Information:

Conference Location: Dublin, Ireland

References

References is not available for this document.