Reputation Analysis of Sensors’ Trust Within Tabu Search

Abstract

Reputation of the sensors are highly essential in authenticating the sensing data, especially when sensors are deployed in a hostile environment. We refer to the term data-trust as the degree of confidence, which can be represented as a quantitative score, based on the reputation of the sensor, where the reputation is comprehended with spatial and temporal redundancy. In this paper, we analyzed the vulnerability of the sensors subject to radical environmental conditions, and we have derived a first-order differential equation utilizing a linear combination of trust factors to quantify the trust value. The trust value is expressed as a weighted combination of two trust factors: coherent data (spatial redundancy) and periodic behavior (temporal redundancy) of the sensors. The selection of weights are automated based on a cost value suited for the operating environment, and it is treated as a combinatorial optimization problem, with an objective function is to maximize the confidence of the sensor. We employed Tabu Search to find the better combination of weights to be associated with the trust factors, in order to find the positive subspace reflecting the domain of trusted sensor operations. We carried out many experiments with varying proportions of the selected trust factors; and the experimental outcomes were analyzed in drawing boundaries of trusted domain (trust space). Our experimental results with varying malfunctioning sensor readings showed that Tabu Search reduced the search space by 22% in comparison to the local search utilizing Simulated Annealing.