Abstract
In this paper we present a distributed architecture to address the problems of managing, tracking, and predicting astray person in large crowds. Our case study considers the pilgrimage to Mecca (Hajj) since it is the largest single annual gathering worldwide. The pilgrimage, lasts for five days and consists of visits to and overnight stays in holy sites in and around Mecca. The path and times of the visits are predetermined. The proposed architecture consists of a user side which is a wearable device that is battery powered and a centralized server side that has a database with the identity information and the geolocations of a pilgrim during the whole pilgrimage journey. The wearable device has a global positioning system (GPS) module that continuously reports the geolocation of the pilgrim to the centralized server. The server contains a map of the whole pilgrimage to correlate the current time and location to where the pilgrim is expected to be in space and time. This arrangement is used as the basis of prediction of the astray status. A list of astray pilgrims is maintained by the system and appropriate messages are sent to the authorities with the location and identity information of the pilgrim. The system also generates an alarm forecasting potential deviations from the prescribed sequence of events and visits to holy sites that could invalidate the pilgrimage. This work also shows a high-level design of the wearable device. The device employs a low-power consumption algorithm to extend battery life for the duration of the journey (at least five days). The server-side database is being implemented using the Neo4j graph database since it is our belief that it is more efficient and compact as compared to the relational database due to the number of joins required for astray prediction queries.