Probabilistic Threshold Based Monitoring Using Sensor Networks

Detecting abnormal events in a monitored area is one of the fundamental applications in Wireless Sensor Networks (WSNs). Accidents and property damage can be avoided if accurate alarms are informed on time. In traditional monitoring strategies, a predefined threshold is given and an alarm is triggered when the sensor reading exceeds this threshold. This Single Threshold based Monitoring (STM) suffers from the inferior quality of sensed data, resulting in many false alarms. This paper proposes a Probabilistic Threshold based Monitoring (PTM) method for WSNs, where an alarm is triggered if the probability of the monitored value being larger than a predefined threshold α is larger than τ. The tight upper bounds of the probability that monitored value is larger than the specified threshold are provided. According to the bounds, probabilistic threshold based algorithms for aggregation monitoring are proposed. Extensive performance evaluation demonstrate the effectiveness of the proposed algorithms. By an extensive experimental evaluation using real dataset, the proposed algorithms outperform the STM method in term of false alarm rate.