Optimized Charging Algorithm based on Q-learning Combining with Node Sleep Mechanism in WRSN

Omyia Mubarak Ahmed Albashir

2020-2023

Master in Electrical Engineering

ABSTRACT

In wireless rechargeable sensor networks (WRSNs), a mobile charger (MC) employs to wirelessly transmit power to sensor nodes. In such a context, designing a charging strategy that optimally prolongs the network lifetime is challenging. This thesis presents a novel charging technique for Wireless Rechargeable Sensor Networks (WRSNs). The first major contribution of this work is the development of an MC charging algorithm that considers target coverage, connectivity, reduce the energy consumption of sensor and prolong the life time of WRSN. To this end, an on-demand charging scheme is introduced that optimizes the WRSN's lifetime. Additionally, the Q-learning approach is utilized to maximize the network's lifetime, resulting in the proposed algorithm being called Q-charging Combining with Sleep node Mechanism (QCWSM). QCWSM prioritizes sensor nodes that play a critical role in the network, allowing it to identify an appropriate charging position that fully powers these priority sensors. Moreover, to prevent sensor failures, a charging model is proposed that schedules sensor nodes whose energy levels are insufficient to keep them alive to go into sleep mode before charging. The proposed algorithm is evaluated against previous works and is shown to outperform existing on-demand algorithms by a considerable margin. Additionally, the QCWSM algorithm can adapt to variations in the network topology, making it more robust and versatile. Overall, the results of this work demonstrate the effectiveness of the proposed algorithm in extending the lifetime of WRSNs by efficiently managing the charging process and prioritizing critical sensor nodes.

Keywords: WRSN, Q-learning, on-demand charging algorithm, target coverage, connectivity, sleeping mechanism