We propose a novel approach to event boundary detection, where autonomous agents are deployed in order to minimize the number of transmissions required to discover an event boundary. The goal of our algorithm is to reduce the number of non-boundary node transmissions (i.e. nodes within the event area and not within transmission distance to the boundary), since the sensory data from these nodes are not required for event boundary detection. The algorithm works by first randomly generating a fraction of agents within the event nodes, then discovering and mapping the boundary, and finally reporting the aggregated results to the user. Simulations demonstrate that the algorithm exhibits O(n) efficiency relationship with the event area, which is an improvement over existing methods that show O(n²) relationships. Furthermore, we demonstrate that the boundary of an event may be successfully mapped using the proposed algorithm.
Local event boundary detection with unreliable sensors: Analysis of the majority vote scheme
