CSIR has a long track record in the field of perimeter protection and its intelligent, multi- infrared beam ACL system is well known and respected in the industry. Its latest innovation, christened ‘Neuraloptics’ by the exclusive representative Mzn Sys-Tech International (Mzn), is set to revolutionise intelligent perimeter security, both commercial and domestic, and could go a long way towards replacing the unsightly electrified and barbed wire fences which provide no real deterrent to a determined criminal.
The Neuraloptics system is based on a simple fibre-optic cable and, whereas this medium has been used by other intrusion detection systems, Neuraloptics differentiates itself completely as it makes use of artificial intelligence in the form of neural networks. Notwithstanding this capability the system is simplicity itself to install. A standard Kevlar-coated multimode optical fibre is run along the primary protection, be it a chain link, weld mesh, palisade, brick or concrete structure. It can even complement the existing ACL system, being deployed in the ground on the inside of the perimeter, where it will pick up the unlikely way of beating the ACL through jumping over its 4 m height. If the perimeter has been constructed from different types of fencing and walls this is easily accommodated by tailoring the classification algorithms for the specific structure within the defined loop (typically a 100 m section).
The optical fibre has a laser diode at one end as a transmitter and a simple photodiode as a detector. The light transmitted by the fibre is very sensitive to vibrations, this characteristic having been used to detect events through the change in signal amplitude. Detection is however a long way off from the detection and identification offered by the Neuraloptics system.
In operation any interference with the fence or wall results in a time and amplitude modulated signal from the detector which is transmitted back to the computer in the central control room for processing and classification. To make the system even more user friendly the actual event itself, when classified, triggers the display of an icon (in the form of a photographic image) on the PC screen. Examples of the displays include a person using wire cutters or a hacksaw, a man climbing a fence and even a child playfully running a bar along the chains of the fence. By the way, by using a standard Pentium-based PC the whole process from event occurrence to classification and alarm status display takes less than 500 ms, truly in the 'blink of an eye'.
False alarms are the bane of any intrusion detection system, as too many of these and personnel tend to ignore all alarms. The ACL system has demonstrated the power of the neural network in minimising this problem, and in one 26 km installed perimeter protection system (with almost 200 zones) out of 14 000 events which occurred in a 24 h period only some 15 were classified after processing as real. Remember that with the ACL, which is an 'open' system, real events include animals and birds, which are classified as such. Raymond Pretorius, the MD of Mzn, conservatively estimates that Neuraloptics will exceed 95% certainty in regard to the classification of alarms as real, false or nuisance. Neels Engelbrecht, the inventor of both the ACL and Neuraloptics systems believes that the classifier of the new system is orders of magnitude better than that of the acclaimed ACL.
A major advantage of the Neuraloptics system is of course that it is totally free from EMI problems and lightening strikes do not affect an optical fibre as it is non conducting. Strikes could activate the detector through the sound/pressure wave, but again this can be classified as not requiring reaction, as are the effects caused by strong winds. As power drawn by the system is minimal, a battery back-up or UPS on the computer will ensure that the system remains fully operational even in the event of a complete power outage - of course cameras and other additional electronic security systems will probably be down.
If tuned to maximum sensitivity and with the correct algorithms the Neuraloptics can actually identify the size of hacksaw being used to cut through a metal fence, as each blade has a distinctive characteristic which depends on teeth size. Neural network-based systems are of course continuously upgradeable, and the computer can be taught new tricks that might be attempted by intruders. The system in theory could even be programmed to detect Buck Rogers trying to cut through that wall with his laser ray gun, although generation of a real icon might cause some problems for Mzn.
Finally, while ACL uses an infrared beam, which can be attenuated in adverse weather conditions, the Neuraloptics installation is totally immune to all types of weather, including fog and heavy rain. As with the ACL system the management software includes a mimic map of the protected site that displays the perimeter zones and the individual zone alarm status. This allows the operator to identify both the alarm and its exact location.
As for Mzn itself, the company was founded in the late nineties by a group of ex Armscor managers and engineers who had been active in the foreign trade area. Using their network of overseas contacts they have been very successful in both project management and marketing in regions which include the Middle East. CSIR's ACL has also proved successful in the international arena with some huge installations, and Mzn took over the marketing rights for this and other related perimeter protection products from CSIR during late 1999.
Neuraloptics will be launched officially by Mzn Sys-Tech International early next year. Interested parties should contact Raymond Pretorius or Danie van der Walt on (012) 460 5578.
© Technews Publishing (Pty) Ltd | All Rights Reserved