Perimeter Security

Some of the earliest work on fiber optic distributed acoustic sensing suggested perimeter security as an application. Sensing cables are buried along boundaries, typically 0.2–0.4 m deep, to detect footsteps or vehicles approaching; the fibres can also be included within fences to alarm on signs of people climbing over, or cutting, them.

In practice, however, it is not sufficient to raise an alarm when the local condition of the fibre changes, indicating a disturbance. The need for intrusion detection is better stated as having a high probability of detecting one of the event types that are pre-defined as threats whilst having a very low probability of alarming on conditions that are not threats. The system must be able not only to detect an event characterized by a change in the backscatter signal but also apply higher levels of processing to discriminate between benign events and genuine threats.

The problem is, of course, more challenging in fibre sections that are habitually noisy, for example, bordering roads or close to built-up areas: the events that precede a threat might be the approach of a vehicle, followed by people walking near the asset to be protected and finally digging. In populated areas, there are many innocent explanations for signals such as these. In addition, the increased ambient noise from non-threatening events makes detecting and classifying real threats much more difficult, especially with a system with random transfer functions, such as amplitude DVS.