Quantitative models for friendly jammer trustworthiness evaluation in IoT networks

Abstract

Due to the open nature of the wireless communication in an Internet of Things (IoT) network, transmitted messages are easy to eavesdrop. To address the eavesdropping attack, we consider an IoT-based smart home where an IoT device (IoT-D) is selected as a friendly jammer to send artificial noise (AN) to protect transmitted messages from being eavesdropped. However, the selected IoT-D may send partial (or even none) AN, which is called an untrusted jammer. To avoid selecting an untrusted jammer, we aim to design two quantitative models from the perspective of time and space to evaluate the trustworthiness of IoT-Ds. Firstly, we adopt an improved energy detector to better counter signal-to-noise ratio (SNR) fluctuation of AN detection. Secondly, based on the detection results, we adopt two trust models: (1) the spatial-oriented beta trust model, (2) the time-oriented hidden Markov model (HMM)-based trust model. In the first model, direct and indirect trust of an IoT-D are both calculated to obtain an integrated trust value. In the second model, the probability of trusted state is calculated as the trust value, and time intervals between detection results of an IoT-D are taken into consideration. Numerical results are presented to demonstrate the performance of our improved energy detection method and two trust models.