Nabil Sarhan, Ph.D., assistant professor of electrical and computer engineering in WSU's College of Engineering, has received $290,000 from the National Science Foundation to develop a security system that performs computer-automated, real-time threat detection at a lower cost and with greater coverage than today's security systems. Once the system is completed, it will be deployed in the College of Engineering for testing.
Currently, video surveillance is primarily used as a deterrent for theft and a forensic tool. Some large organizations have systems that detect suspicious activities as they occur, using anywhere from hundreds to millions of cameras that are observed at a central location by trained personnel. These systems are less than ideal, however, because they are subject to human error and typically cost millions of dollars.
To amend these problems, Sarhan's project considers the design of computer systems that automatically monitor and assess potential threats.
"The problem with having people monitor large numbers of screens all at once is that humans are unable to pay close attention to multiple events that happen at the same time," Sarhan said. "A computer system, on the other hand, can efficiently process a large number of surveillance videos simultaneously, without getting tired, bored or losing focus."
The major benefit of Sarhan's design is that it will dramatically lower the price and increase the coverage of computer-automated security systems. Large scale systems are too costly for many businesses because they require extensive amounts of bandwidth and energy to run a large number of cameras that produce and transmit high quality surveillance videos. In addition, powerful computers are needed to simultaneously process the information fed from the system's camera.
To achieve a more cost-effective use of bandwidth and battery power, Sarhan is developing dynamic "bandwidth allocation schemes" which prioritize video streams based on the security level of the camera location, current threat level and distance from other high security areas. This means less energy and resources will be distributed to a camera in low-security areas where no threats are detected than those in a high security area with several threats nearby. This dynamic system will automatically adjust with any changes in a location's threat level. An additional feature will also determine the current overall threat level of an entire site based on all of the risk factors together.
Sarhan's system will allow the use of less-advanced computers to process the information transmitted from the system's cameras, further lowering the cost of owning a system.
Sarhan's strategy for making security systems more efficient stemmed from his research in optimizing multimedia systems. One of his projects uses similar concepts of using resources more efficiently to make Internet video streams clearer and have fewer interruptions.
"We're applying a lot of similar principles to Internet video streaming as we are to this security system, such as the idea of being smarter about how we use our resources," Sarhan said. "To be using this knowledge for security systems, however, feels like a more worthwhile pursuit. The best science is the kind that contributes to the well-being, safety and security of humans and their properties. That is what we are trying to do here with this project."
E-Mail Story
Print Story
