Energy-efficient node position identification through payoff matrix and variability analysis

Applications for mobile ad-hoc networks are heavily dependent on information about the nodes positions. Since the network may include malicious nodes providing bogus data (e.g., fake positions), the reliability of this information is critical. Although this problem has been addressed by some predictive models, challenges still exist regarding (a) the accuracy/security of such models, (b) the potential advantages of combining different prediction models, (c) the power consumption resulting from the simultaneous application of multiple models, and (d) the lack of strategic approaches in the analysis of information aiming to find the most advantageous data and balance divergent results. To address these issues, we analyze in this paper the performance of two prediction methods: linear regression and Grey model. We have evaluated their corresponding energy costs for reliably identifying each node position. The results, obtained via simulations, are added to independent vectors and statistical indicators are gathered to create a game theory payoff matrix. The proposed model allows the evaluation of the predictive methods either individually or collectively, facilitating the identification of the best parameters for a target energy saving profile.