Modeling the social internet of things network using random graphs

Abstract Social Internet of Things (SIOT) is a paradigm in which thing communicate with one another based on social relation. The goals is for the things to search for services, retrieve and provide users with them independent from their owners. The existing approaches lack models for analyzing the efficiency of SIOT. The objective of the current research is to model SIOT with regard to various relationships and their different topological properties. In this approach, the graphs related to SIOT are modeled based on similarity of their topological properties and random graphs in such a way that these properties are preserved by increasing the size of the network, the intended topological properties are preserved. For this purpose, first, topological properties of real SIOT graphs have been extracted. Then, using numerical and intuitive comparisons, the degree of resemblance between SIOT topological properties and random graphs has been examined. In order to prove this resemblance and network scalability, the connection between average route length and descending gradient algorithm has been implemented. The obtained results have shown the resemblance of ownership object relationship (OOR) real graph to Erdos Renyi (ER) random graph per p = 0.9, parental object relationship (POR) graph to ER random graph per p = 0.009, co-location object relationship (CLOR) graph to ER random graph per p = 0.00009 and social object relationship (SOR) graph to Barbasi Albert (BA) random graph per m = 50. In order to evaluate the proposed framework, the real SIOT dataset has been used and the scalability and maintaining the topological properties have been proven.