Energy-Efficient and Fault-Tolerant Evolution Models Based on Link Prediction for Large-Scale Wireless Sensor Networks
Shuaizong Si,Jinfa Wang,Chong Yu,Hai Zhao
DOI: https://doi.org/10.1109/access.2018.2882389
IF: 3.9
2018-01-01
IEEE Access
Abstract:Wireless sensor networks (WSNs) deployed in harsh environments, i.e., battlefield and natural disasters areas, often suffer from the problems of the deliberate attack, hardware failure, and energy depletion. It is crucial to propose the fault-tolerant and energy-efficient evolution models to accomplish tasks of applications in these environments. Previous studies have reported that scale-free (SF) topologies can improve the fault-tolerance of WSNs. However, energy-efficiency is less concerned in most of these SF WSNs models. In addition, previous models took only node degree into account in calculating connection probabilities between two nodes, topological characteristics, which can enhance reliable communication, are rarely considered. To address these shortcomings, we present three new SF-evolution mechanisms (ECN, ELP, and ELCP) of large-scale WSNs from the novel aspect of link prediction. Specifically, three well-known link prediction methods, which are topology concerned, named common neighbors (CN), local path (LP), and local community paradigm (LCP) are adopted in the corresponding models to enhance energy-efficiency of SF-WSNs. Meanwhile, residual energy and degree capacity are taken into consideration in the proposed models to balance the node energy consumption. We also explore the degree distribution of networks generated by our models and present the theoretical derivation of it. The experimental results demonstrate that our proposed models can generate SF-WSNs topologies with better fault-tolerance and higher energy-efficiency by comparing with a candidate clustering-based algorithm and other two SF enhancing algorithms. Moreover, ELCP’s performance is the best among the three models, followed by ELP and ECN.