By Tom Condon, RPA, FMA
From the August 2013 issue of Today’s Facility Manager
Some facility managers (fms) have specific security requirements in that they need to protect large perimeters around their facilities. This is common in airports and other government facilities, as well as manufacturing and some large corporate campuses that require high security. Perimeter intrusion detection systems, or PIDS, have many challenges not often encountered inside buildings. Weather and animals can cause false alarms and can damage components, and perimeters typically are very large, which often results in significant expense to secure them.
Perhaps the oldest and simplest of perimeter detection is the strain sensor, which uses a taut wire attached to a fence and connected to a device that measures how much the wire is pulled (when someone moves the fence by climbing over it or cutting the wire). While low-cost and relatively simple, these can only be attached to certain types of fences, require regular maintenance, and are affected by lightning.
The latest evolution of fence mounted sensors is fiber optic cable that detects movement by sensing the changes in the way light travels inside the cable. Movement alters the characteristics of the light signal, and these signal changes can be analyzed to detect intrusion while ruling out movement caused by wind or rain. Also, these cables are not subject to the challenges of taut wire systems noted above.
Another basic technology that has been around for many years is the use of photoelectric beams extending from a transmitter to a receiver. These are low-cost and easy to install, but are subject to a fairly high rate of false alarms from animals, rain, fog, etc. The limitations of photoelectric beams have prompted vendors to introduce new systems in a similar configuration, but they use microwaves instead of light. This provides a larger field of detection than a single beam of light, has a much lower rate of false alarms, performs better in harsh weather, and can provide a “virtual wall” of intrusion detection up to 650′ long.
One of the more sophisticated PIDS uses a pair of parallel cables positioned a few feet apart. One of the cables broadcasts radio frequency signals, and the other cable detects the electromagnetic field created by the radio signals. This forms a sensor field that extends about 3′ around the cable pair (multiple pairs of cables can be grouped to form larger fields). These systems can detect an intruder passing through the field, and they are also capable of detecting an intruder’s size because greater body mass affects the field more. This enables the system to differentiate between a person and a wandering rabbit. These cables can be strung between posts above ground (usually in an array 6′ to 8′ high, which can provide a detection field up to 18′ high), or can be buried underground for an invisible detection field.
To enhance a PIDS, many facilities will link the alarm outputs of the perimeter system to the video surveillance system. When an alarm occurs, the PIDS sends the location of the alarm to the video management system, which then moves the nearest pan-tilt-zoom (PTZ) video camera(s) to zero in on the alarm site. This allows operators to immediately see what is happening at that site and is often crucial in detecting false alarms.
The latest PIDS technology centers around video analytic systems that use video cameras and intelligent analytics to detect intruders. Cameras are positioned along the perimeter, and their video output is analyzed to detect changes in the pixels of the image. These changes are assessed by specialized software designed to understand that a group of pixels moving together is an object, person, or animal. The analytics software can then determine where the threat is located, which direction it is moving, and how fast it is moving. Some software is so sophisticated that it can discern between a person and a similar sized animal based on its shape and movement. Using cameras alone can have significantly lower costs than installing fences, digging trenches, and running cables, because a single high-quality camera can cover a very large area. But keep in mind that some weather phenomenon, like fog, can be challenging for video cameras, so thermal cameras may need to be installed.
Video analytics software has some very powerful features, including the ability to draw virtual fences around areas in a camera’s view. To configure this, the operator uses a computer mouse to draw lines around protected areas in an on-screen camera view. Once a protected zone is created, users can choose settings that tell the software to alert them when it sees movement. This allows one analytic camera to view protected and non-protected areas, and to know the difference between them.
Analytics can do amazing things, but it requires significant amounts of computer processing power, so these systems employ pretty beefy servers. In an effort to ease the load on servers and control costs, some vendors have come up with a way to handle the load of analyzing video: by moving processing power to the “edge.” This means moving power into the cameras themselves. By building computer microprocessors into cameras, video analytics can be performed inside the camera, thereby distributing the load across many smaller processors at the “edge” rather than a small number of high-powered servers. As the cost of computer chips continues to drop, this strategy will be used more.
For a reference guide on both indoor and outdoor intrusion detection systems, the Nuclear Regulatory Commission (NRC) has a document available as a pdf download here.