Research Article
Spectrum-Agile Cognitive Interference Avoidance Through Deep Reinforcement Learning
@INPROCEEDINGS{10.1007/978-3-030-25748-4_17, author={Mohamed Aref and Sudharman Jayaweera}, title={Spectrum-Agile Cognitive Interference Avoidance Through Deep Reinforcement Learning}, proceedings={Cognitive Radio-Oriented Wireless Networks. 14th EAI International Conference, CrownCom 2019, Poznan, Poland, June 11--12, 2019, Proceedings}, proceedings_a={CROWNCOM}, year={2019}, month={8}, keywords={Deep Q-network Deep reinforcement learning Interference avoidance Wideband autonomous cognitive radios}, doi={10.1007/978-3-030-25748-4_17} }- Mohamed Aref
Sudharman Jayaweera
Year: 2019
Spectrum-Agile Cognitive Interference Avoidance Through Deep Reinforcement Learning
CROWNCOM
Springer
DOI: 10.1007/978-3-030-25748-4_17
Abstract
This work introduces a spectrum-agile wideband autonomous cognitive radio (WACR) that is capable of avoiding interference and jamming signals. Proposed cognitive technique is based on deep reinforcement learning (DRL) that uses a double deep Q-network (DDQN). Moreover, it introduces new definitions for the state and the operation parameters that enable the WACR to collect information about the RF spectrum of interest in both time and frequency domains. The simulation results show that the proposed technique can efficiently learn an effective strategy to avoid harmful signals in a wideband partially observable environment. Furthermore, the experiments on an over-the-air channel inside a laboratory show that the proposed algorithm can rapidly adapt to sudden changes in the surrounding RF environment making it suitable for real-time applications.