Sensor Single and Multiple Anomaly Detection in Wireless Sensor Networks for Healthcare

Abstract

Wireless Body Area Network is used in healthcare applications for collecting information from remotely monitored patients. As they work in medical Applications, these kinds of networks must be robust and flexible to the sensors failure. This means that it must differentiate between patient's emergency alarms, and ill-behaved sensor's false alarms. In this paper, we propose an approach for faulty measurements detection in order to make alarming of emergency situations more precisely. The proposed approach is based on decision tree, threshold biasing and linear regression. Our objective is to detect single and multiple faults in order to reduce unnecessary healthcare intervention. The proposed approach has been applied to real healthcare dataset. Experimental results demonstrate the effectiveness of the proposed approach in achieving high Detection Rate and low False Positive Rate. The ability of this algorithm to detect single and multiple anomalies make it more reliable for medical emergency use.