Evolutionary game model based on cumulative prospect theory for information management mechanism in SIoT

As nodes in Social Internet of Things (SIoT) become more intelligent, malicious information occurs more frequently and spreads more widely. This problem can severely affect the trustworthiness of services and applications in SIoT. Methods to effectively control malicious information spreading in SIoT are essential and necessary. Reputation mechanism provides a powerful tool to tackle this challenge. In this paper, we propose a reputation-based mechanism to activate the self-purification capacity of the SIoT network by balancing information conflicts triggered by reporters and supporters. In order to find the best rewarding and punishment strategy, a bilateral cumulative-prospect-based evolutionary game model of SIoT network information conflict is constructed. Using local stability analysis and numerical simulation, the evolutionary trends of the proposed game model under different theoretical application scenarios are analyzed. The findings indicate that the basic income and deposit of both sides, the popularity of information as well as the importance of the conformity effect all have a significant impact on the system's steady state and evolutionary path. The specific conditions that both participating sides of the game tend to treat conflicts relatively rationally are analyzed. Dynamic evolution analysis and sensitivity analysis of selected parameters show that basic income is positively related to smart object's feedback strategies, while deposit is negatively related to that. While weight of conformity effect or the information popularity goes up, the rising of feedback probability is observed. Based on the above results, suggestions on dynamic reward and punishment strategies are given. The proposed model is a helpful attempt to model the evolution of information spreading in SIoT networks, with the ability to simulate several well-known regularities of message dissemination. Proposed model and suggested quantitative strategies can be helpful to build feasible malicious information control facilities in SIoT networks.