Dynamic Trustworthiness Overlapping Community Discovery in Mobile Internet of Things

Abstract

In the mobile Internet of Things (IoT), dynamic community discovery has attracted increasing attention, because it can provide strong reliability and security for mobile IoT applications, such as data forwarding or fusion. However, when building community structures, most traditional global detection methods only focus on the encounter ratio between nodes and need to have holistic cognition for the whole network structure information. Thus, in IoT applications with large-scale or incomplete structures, there are deficiencies, such as lower security and higher time complexity. This paper proposes a detection scheme for dynamic trustworthiness overlapping community, called D2-TOC. This scheme first employs evidence-based data between node pairs, such as contact probability, recency, and service degree, to construct the trustworthiness relationships between devices and the network model of mobile IoT, which can provide the security guarantee for data interaction from the start. Second, the D2-TOC uses optimized random walks and seeds expansion methods to fulfill the initialization and evolution of overlapping communities, which avoids the high temporal and spatial complexity generated by global community detection. Finally, the experimental results have proven that, in mobile IoT scenarios, our D2-TOC algorithm has a better performance than the DEMON, Kim-Han, LFM, and UEOC in accuracy and efficiency.

Keywords

• Communities
• detection algorithms
• Internet of Things
• random walks
• seeds expansion
• trustworthiness