Fuzzy-Layered Recurrent Neural Network Based Hybrid SWIPT Protocol for Cooperative Networks

In this letter, we present a novel Fuzzy-Layered Recurrent Neural Network (Fuzzy-LRNN) model for optimization of hybrid Simultaneous Wireless Information and Power Transfer (SWIPT) in a cooperative communication scenario. At first, the Fuzzy rule set has been devised to obtain the initial estimates of time switching factor and power splitting, considering the source transmit power and channel conditions of both hops. Afterwards, these estimates are then fed to an LRNN, whose learning is carried out through a dynamic (optimal) model, for compensating the initial estimates to near-optimal values. The performance of Fuzzy-LRNN approach is evaluated in terms of end-to-end outage and the comparison is carried out with Time Switching (TS), Power Splitting (PS), Hybrid Protocol (HP), dynamic, Fuzzy only and LRNN only schemes. The results indicate that the proposed Fuzzy-LRNN with a lower complexity compared to LRNN can achieve a better outage performance and reaches the outage results of a dynamic approach.