Identification of Bee Colony Acoustic Signals using the Dynamic Time Warping Algorithm

The work deals with automated recognition of the current state of a bee colony, for continuous monitoring of processes running in a bee hive is of key importance in beekeeping. The dynamic time warping algorithm is considered as a method of analyzing acoustic signals produced by a bee colony. Upon such an analysis one can make inferences about the current state of the colony. We have developed a software module for audio-signal identification, which is to be used as a part of an automated bee colony monitoring system, and a software tool for verification of the module. We evaluated the efficacy of the algorithm, the probability of bee colony states correctly recognized using acoustic signals produced by the colony and consumed computational resources by the example of a queen bee’s sounds recorded during swarming. The dependencies of the signal processing time and the successful pattern recognition probabilities on the frame sample rate and frame size are presented.