Remote physiological monitoring of first responders with intermittent network connectivity

First responders have been observed to be at increased risk of cardio-vascular diseases compared to the general population. A high percentage of cardiac events have been found to occur during missions. Continuous physiological monitoring during missions can be effective in reducing the number of fatalities. Real-time physiological data such as ECG can be collected using body-worn sensors. This sensor data can be processed on the body itself or can be communicated over an ad hoc wireless network to the incident command center located nearby. First responder missions often take place inside building structures where network connectivity is intermittent. Intermittent connectivity can lead to loss of critical physiological data or delay in that information reaching the base station. Hence, some amount of local processing is needed in order to limit the amount of data that is communicated. In this paper, we introduce a novel Hidden Markov Model based classifier for myocardial infarction detection. The classifier fidelity can be adapted based on the processing power available. We present a peer-to-peer networking protocol for communication over disrupted networks. A low fidelity classifier is used to perform local processing and assign priorities to the data based on its criticality. It is complemented by a disruption-aware epidemic forwarding protocol for transferring first responder's physiological data to the base station. We show that with prioritized epidemic forwarding and buffer eviction policy, packet delivery ratio for abnormal data increases and the latency associated with abnormal packets reaching the destination decreases.