Abstract: A computer-implemented event statistic generation method for intrusion detection comprises processing a plurality of return signals from a coherent optical time domain reflectometer into time domain signals for each of a plurality of sensor bins, the plurality of return signals corresponding to a plurality of stimulation pulses injected into an optical sensor fiber during a time period.

Abstract: A computer-implemented event statistic generation for intrusion detection comprises processing a plurality of return signals from a coherent optical time-domain reflectometer into time-domain signals for each of a plurality of sensor bins, the plurality of return signals corresponding to a plurality of stimulation pulses injected into an optical sensor fiber during a time period. For each sensor bin, transforming the respective time-domain signal into a corresponding frequency-domain signal, calculating, from the respective frequency-domain signal, a first signal power area of a first frequency band expected to contain system noise and a second signal power area of a second frequency band expected to contain any energy related to at least a first event; and generating an event statistic proportional to the ratio of the second signal power area to the first signal power area at least in part by dividing the second signal power area by the first signal power area.