An Adaptive Multipath Routing Method Based on Improved GA and Information Entropy
Jianhang Liu, Xu Meng, Shibao Li, Xuerong Cui, Haibo Wu · IEEE Sensors Journal · 2022
With the development of artificial intelligence technology, some emerging applications (such as computation offloading and digital twin) require higher transmission quantity and quality for wireless sensor networks (WSNs). The increasing amount of data aggravates the imbalance of node energy consumption. In the single-path routing, nodes on the optimal path consume too much energy, which leads to the premature failure of the network. In addition, the transmission protocols based on the acknowledge character (ACK) mechanism ensure the quality of service, but bring additional energy overhead. In this article, an adaptive multipath routing method based on balanced energy consumption (AMRBEC) is proposed for WSNs that significantly reduce energy consumption and prolong network lifetime. We design a new multidimensional fitness function based on the path energy, distance, hops and the lowest energy node, which servers to an improved genetic algorithm (GA) to select the multiple optimal paths. The packet loss rate is predicted by random forest to realize the non-feedback transmission. According to the residual energy and the information entropy of each path, the amount of data transmitted is dynamically adjusted, and the optimal paths are updated to balance the energy consumption of the whole network. A simulation analysis shows that the proposed AMRBEC performs better than other algorithms from the literature by improving the lifetime by 20% and reducing energy consumption by more than 20%.