Flexible And Functional
Studies have established that human operators cannot effectively monitor video from multiple security cameras over long durations. Video analytics, also known as intelligent video, addresses this issue by analyzing incoming video footage for suspicious behaviors and alerting the appropriate personnel to such events. Analytics also plays an important role in the scalability of video surveillance networks. As camera networks grow to keep pace with expanding government facilities and multiplying security threats, cost-efficient ways to manage data traffic and video storage requirements become important considerations for security directors and IT personnel.
Choosing the right system
Intelligent video takes two forms: edge devices and server-based solutions. Edge devices, named because of their location on the outside boundaries of a network, include smart cameras and encoders that convert analog video to IP. They act as input devices to the main video management system and feature specialized onboard processors to analyze behaviors at their point of capture. Because of the limited onboard processing capacity, these devices support only a single device or fewer than four cameras and offer a limited set of analytics capabilities.
In contrast, server-based video analytics systems function parallel to or act as the primary video management system, processing and analyzing input from larger numbers of cameras at a centralized location. Server-based systems also tend to offer a more robust analytics capability and can detect multiple concurrent events or behaviors.
With both options available, a debate is under way about whether distributed edge devices or centralized server-based solutions are better suited to scale for large, enterprise-wide video surveillance systems. Security directors need to weigh the importance of several factors inherent with each approach including price, maintenance, operational flexibility and bandwidth.
Edge products, a newer technology, tend to have higher price points, which can designate wide-scale implementations as “budget breakers.” More often than not, edge devices are implemented in small numbers or only in select locations, to cover a small area within an agency’s operation. Server-based systems, an older and proven technology, can cost far less per video channel and can be inexpensively deployed to encompass entire facilities.
Regardless of the selected solution, it is important to consider every cost component, including hidden costs, in the total equation. Cost components include:
cost/behavior/camera (edge) versus cost/behavior/channel and number of channels necessary to support all cameras (server-based);
cost of limited event detection and incremental number of edge devices necessary for sufficient coverage (edge);
incremental hardware costs if the intelligent video system is software-only (server-based);
engineering costs to install and “tune” a system (both types);
operator training costs relative to the annual turnover rate (both types);
network upgrade costs necessary to ensure burst-free, real-time transmission of video alerts (both types);
security camera costs to upgrade video quality (server-based) or replace existing cameras (edge); and
“commitment” cost if the proprietary system only integrates with a limited number of components (both types).
System maintenance and setup are important components to an effective intelligent video installation. Yet, they are often points that are overlooked by government security directors who are looking to “beef” up video systems immediately. Failure to recognize the differences between each solution, however, will lead to future problems, as agency facilities grow and require expanded video analytics coverage.
Both server-based and edge devices suffer if they are based on proprietary custom hardware. Installation becomes a marathon that requires calibration of each device installed on the network. Also, because the analytics component is built into the device’s firmware, upgrading to a newer version or applying a bug fix requires almost as much work as the initial implementation.
Many server-based analytics operate on commercial-off-the-shelf (COTS) hardware. Agency IT personnel can easily conduct the installation themselves as if it is just another addition to the network infrastructure. Configuration of individual cameras and channels can be done from a central location without having to manually work on each device involved. The software is also easily updated, either through server-side installations or IP-enabled patching routines initiated by the developer.
Is it easier to upgrade a server or to install a new set of security cameras? Almost every agency security director and IT director will lean toward the server option because of its increased flexibility. New additions to the camera network can be plugged directly into the existing network infrastructure and added to existing channels. Furthermore, centralized server-side analytics integrate into the video management system and offer unlimited scalability when multiple customers wish to view or search video.
Edge devices require more effort to scale as a sole solution for large video surveillance systems. However, they can be an important and effective element of an enterprise-wide system and should be considered supplementary to a large video surveillance architecture.
From a bandwidth perspective, edge devices beat out server-based systems because of their unique position on the network. As smart cameras and video encoders process images at the network edge, they only record and transmit notable or pre-set events, thus reducing the data costs related to transport. Server-based solutions tend to be more centralized within the network and require the information be sent in for processing and then out again for viewing, eating up more bandwidth.
No matter which technology is selected, both edge devices and server-based systems reduce storage costs associated with video management systems. Because video analytics software tags data as it is processed, only notable footage is saved. This reduces hundreds of hours of footage transmitted from several hundred cameras to a fractional amount of relevant video on daily basis.
With this information in hand, which solution is best for a government security director who is seeking a scalable solution for his or her facility? Ideally, that which best supports the ultimate objective – to ensure the safety and security of people and assets. Since it is impossible to predict the exact location of the next security violation, and since human monitoring for long periods of time is often ineffective, it is a flawed and limited approach not to place analytics on every camera. A combination of both edge devices and server-based systems provides the optimum blend of cost savings, flexibility and processing power. For large-scale systems, the ease of deployment, the flexibility of system growth and the ability to run on COTS hardware make server-based solutions that are supplemented by edge devices a great approach.
About the Author
Nikhil Gagvani, Ph.D. is the chief technology officer and vice president of engineering for Cernium Corp., Reston, Va., a provider of real-time intelligent video. Gagvini is responsible for Cernium’s patented Perceptrak system, as well as the development of new product lines based on the patented P-Core analytics engine. He holds Ph.D. and Master of Science degrees in computer engineering from Rutgers University and a Bachelor of Technology degree in aerospace engineering from the Indian Institute of Technology.