Research Article
Reinforcement Learning Based Energy Management (Rl-EM) Algorithm For Green Wireless Sensor
@INPROCEEDINGS{10.4108/eai.7-12-2021.2314541, author={V Mahima and A Chitra}, title={Reinforcement Learning Based Energy Management (Rl-EM) Algorithm For Green Wireless Sensor }, proceedings={Proceedings of the First International Conference on Combinatorial and Optimization, ICCAP 2021, December 7-8 2021, Chennai, India}, publisher={EAI}, proceedings_a={ICCAP}, year={2021}, month={12}, keywords={markov decision process reinforcement learning wireless networks internet of things and energy management}, doi={10.4108/eai.7-12-2021.2314541} }- V Mahima
A Chitra
Year: 2021
Reinforcement Learning Based Energy Management (Rl-EM) Algorithm For Green Wireless Sensor
ICCAP
EAI
DOI: 10.4108/eai.7-12-2021.2314541
Abstract
Green network is basically capable of producing its own energy through renewable energy resources. The sensor nodes serving to collect the physical parameters are equipped with small scale solar panels and tiny wind mills to harvest their own energy. The limitations on battery and availability of renewable energy resource create challenges in continuous monitoring without missing any events. The energy harvesting sensor network requires dynamic adaptation to the time varying behavior of the environment. The reinforcement learning based energy management algorithm named RL-EM is proposed which observes and learns from the environment. The voltage level of the sensor node is considered for designing the algorithm. The RL-EM outperforms existing algorithms with 40% reduced sleep nodes and 60% reduced energy overflow with LEACH protocol. The proposed RL-EM algorithm outperforms the LEACH, SEP-M and ACO algorithms in terms of throughput, sleep nodes, energy overflow, and load distributions.
