Two well-publicized incursions at heavily protected U.S. facilities this summer illustrate the glaring weaknesses that still remain in high-tech perimeter security systems. Although the breaches were ultimately benign, the incidents at JFK Airport in New York and the Y-12 nuclear weapons plant in Tennessee serve as not-so-subtle reminders of the high stakes involved in safeguarding our nation’s critical infrastructure.
Two well-publicized incursions at heavily protected U.S. facilities this summer illustrate the glaring weaknesses that still remain in high-tech perimeter security systems. Although the breaches were ultimately benign, the incidents at JFK Airport in New York and the Y-12 nuclear weapons plant in Tennessee serve as not-so-subtle reminders of the high stakes involved in safeguarding our nation’s critical infrastructure.
Based on published reports, these failures seem to have been caused by multiple independent sensor systems that did not communicate with each other in an intelligent fashion. Had these sensor systems been integrated into a network of systems in which sensors interacted intelligently with other sensors, the security networks could have detected, classified, confirmed, validated and identified the threats. The incursions would have been much less likely to have succeeded.
The first incident occurred in July at the Y-12 facility near Oak Ridge. Anti-nuclear protesters reportedly cut their way through three perimeter fences before reaching the depot’s outside wall. They may have been onsite for 45 minutes before apprehension by a security team, which – according to one published report – wasn’t quite sure what to do with them.
The JFK incident a month later would have been humorous if it weren’t so serious. A stranded jet skier swam up to and climbed over the fence protecting an airport runway. According to news reports, a supposedly state-of-the-art $100 million perimeter sensor system failed to notice him before he crossed the airport grounds on foot, reached a terminal and asked for help.
Let me clarify that I have no specific knowledge of the security systems at either facility other than what has been published. Based on my experiences in this field, however, it is safe to assume that both have multiple independent sensor systems surrounding them along with physical barriers. Typically, these kinds of facilities incorporate various types of full motion video (FMV) and pan, tilt, zoom (PTZ) cameras as well as active and passive sensors for detecting motion, heat, sound and other target features or attributes.
Published reports of the JFK incident have not specified exactly where failure occurred in the expensive perimeter security system, but a detail picked up by the news media in Tennessee provides a clue about what possibly happened in both situations. Y-12 sensors, likely on the fences, did their jobs and triggered alarms. Remarkably, these warnings were reportedly ignored by guard personnel.
My educated guess is that security personnel didn’t respond immediately to the alerts because they had received so many false alarms in the past. Perhaps they thought sensors on the fences had been tripped by squirrels or tree branches. Regardless of the explanation, too many false positives are produced from even the most sophisticated sensor systems. These are dangerous because they create complacency, even with trained guards, in a phenomenon called desensitization.
This problem can be avoided by differentiating false alarms from legitimate threats, which is accomplished by integrating the sensors systems into intelligent networks. When sensor systems communicate with each other, smart information is generated, which ultimately improves communication between the network and security personnel. Perhaps the most frustrating aspect of this situation is that technology exists today to get these individual systems interacting in such a way that false alarms can be reduced to zero.
The goal is to move beyond merely detecting a threat. When multiple types of sensors share data and analysis as a network, a clearer, reliable and robust picture of the threat emerges. Working as an integrated system, these networks can detect, classify, confirm, validate and identify threats. Once the threat has been 100% verified as a truck, person or jet ski at a precise location, the integrated network relays a specific warning to the personnel who are best positioned and equipped to respond appropriately.
A rapidly developing technology called Real Time Analytical Processing (RTAP) makes this possible. RTAP employs advanced algorithms to analyze raw sensor data for classification and identification purposes – in a fraction of a second. When multiple sensor systems are integrated properly, RTAP becomes even more powerful by correlating the spatial-temporal information designed to answer the what, where and when questions.
The most critical advantage of RTAP is its ability to cull meaningful information from the data noise that streams relentlessly at high volumes from the sensor systems. With the right architecture, these advanced algorithms ignore the noise while focusing precious compute power only on the valuable bits of data.
With RTAP incorporated, the sensor networks are set up in a daisy chain so that positive threat analysis at one sensor layer is fused with analyses from other sensor systems. Sensor A determines “X” while Sensor B determines “Y”. When X and Y are fused, the information produces a 100% reliable alarm. In many cases, Sensor A may actually activate Sensors B and C, which work together in a network.
Clearly, time is the most important variable for any early warning system. For this reason, RTAP development now focuses on finding ways to eliminate latency in data processing. One way we are working on this is to physically move the analysis capabilities closer to the sensors – or even embed them in the sensors – so no time is lost in transmitting raw data to a centralized processing unit some distance away.
With this latency removed and RTAP algorithms already capable of performing thousands of analytical calculations in milliseconds, the integrated sensor network can detect, classify, confirm, validate and identify the target in less than a second. This means that information regarding a verified threat along with its location or approach vector can be relayed with 100% confidence to the personnel most capable of handling it.
