Faulty Data Detection in mMTC Based E-health Data Collection Networks

In the fifth generation mobile communication (5G) and beyond, massive machine type communication (mMTC) is regarded as a key technology to support the universal Internet of Things (IoT) devices. As a promising vertical application area of IoT, e-health received many attentions from academic, medical and industry. To provide reliable suggestions for doctor, guaranteeing the authenticity of the sensory data during the data collection procedure is the first and fundamental step. Focusing on the mMTC based e-health data collection scenario, we propose the threshold-based decision faulty sensory data detection (FSDD) scheme by analyzing the correlation among several physiological parameters to guarantee the authenticity of the collected data. In the proposed FSDD scheme, machine learning algorithm is exploited to predict the ground truth value of a certain physiological parameter. By searching the dynamic optimized threshold to optimize the performance of the system, the FSDD scheme provides great performance improvement in high detection accuracy and low false alarm. Furthermore, the proposed FSDD scheme can recognize the deteriorated health condition. The effectiveness of proposed FSDD scheme is verified by simulating on a real medical database.