In wireless rechargeable sensor networks (WRSNs), charging request nodes (RNs) are characterized by several criteria which are contradictory.
Recently, on-demand charging scheduling schemes, which use two or more multicriteria decision-making (MCDM) methods, have been proposed. However, these schemes use a pairwise ratio scale which can magnify the actual pairwise difference between multicriteria, and do not take into account the trade-off between performance metrics.
Ri Man Gun, an institute head at the Faculty of Communications, has proposed an on-demand charging scheduling method based on a fuzzy cognitive network process (FCNP) which uses a fuzzy pairwise interval scale.
The proposed method, called an integrated FCNP-Q-learning-based scheduling (iFQS), first uses FCNP to exactly assign the relative weights to five multicriteria for charging prioritization and to three multicriteria for partial charging time (PCT) determination, respectively.
Then, in charging path planning with Q-learning, the BS uses these five criteria’s weights to design the reward function and select the most suitable next charging sojourn point. On the other hand, the three criteria’s weights are used to reasonably determine the PCT at charging sojourn points while achieving a desirable trade-off between charging metrics.
The results of the extensive simulation show that the iFQS significantly improves charging performance in comparison with the existing MCDM-based methods.
You can find the details in his paper “iFQS: An Integrated FCNP-Q-Learning-Based Scheduling Algorithm for On-Demand Charging in Wireless Rechargeable Sensor Networks” in “International Journal of Distributed Sensor Networks” (SCI).
© 2021 Kim Chaek University of Technology