Measurement and security trust in WSNs: a proximity deviation based approach

Quality of communication and measurement accuracy are of prime importance in WSN-based applications, as the sensors have to report real-time measurements to enable efficient decision-making. These sensors are often subject to measurement errors, such as noise, nonlinearity, and deviation caused by rapid changes. Sensors are prone to failures and can be targeted by attacks that aim to modify their outputs. To address these drawbacks, measurement should be checked, sensor functions should be protected, and accuracy should be analyzed all the time. In this paper, we propose a trust management framework based on three metrics: the true measurement deviation, group deviation, and security metrics to analyze the accuracy and trustworthiness of sensor data. The third metric considers the attacks detectable at the communication architecture layers. We have derived an analytical model (i) to calibrate the sensors periodically by deciding on their trustworthiness, and (ii) to modify their outputs periodically, whenever needed based on the deviation in measurement. Our simulation results for a real-time fire monitoring system show that the proposed trust framework is efficient in producing 95% accurate and trusted measurements by limiting the frequency of sensor calibrations to a minimal value and by setting a lower boundary of 5% deviation from the true and group value metrics with its added trust at the micro-level by disproving attacks at the protocol layer.