Where the camera keeps its brains
Video surveillance systems are becoming more intelligent every day.
By Colin Bodbyl
Some of the world’s largest companies now invest heavily in the software behind the surveillance system, including artificial intelligence and video storage. While development houses forge ahead creating new features and enhancing the capabilities of video surveillance software, the physical location of where that software should operate is not always clear.
There are three locations in a surveillance system where software can operate. The first is on the camera, commonly referred to as “at the edge.” Running software at the edge has some significant benefits. For one, the more software that runs on the camera, the less reliant the system is on a centralized server. This reduces the points of failure and generally results in a more fault-tolerant system. A good example of software services that are beneficial at the edge would be video recording. If a camera is able to record to a storage drive on the camera itself, critical recordings are no longer at risk of being lost to network interruptions or even complete network failures.
The biggest downside to running software at the edge is cost. Cameras with the processing power to run video analytics or house internal storage drives are expensive. While this is typically offset by savings in other areas of the initial price of the system, it can make upgrading the system in the future significantly more expensive. Camera technology is constantly evolving and becomes better every year with the release of higher resolution cameras with better image sensors. Unfortunately if users invest in too much intelligence at the camera level it can put them in an awkward position when those same cameras are quickly outdated and expensive to replace.
The second location where software can be located is at the NVR (or onsite server level). This is arguably the most popular location for software, requiring a lot of processing or storage. Servers can be built with powerful CPUs, able to support software upgrades and changes for years to come. Operating systems like Windows allow users flexibility to change or add new software and in turn features with the click of a mouse. Unfortunately, servers also create a single point of failure. While servers can be backed up and protected from failure using different devices, they are ultimately the brain of the surveillance system and should anything happen to them the entire system stops functioning.
The third and final software location is the cloud. While the cloud is essentially an offsite server, it offers redundancy that no onsite server can match, including guaranteed uptime. The cloud eliminates the need for intelligence on the camera, and can eliminate the need for onsite servers altogether. It also offers the ultimate in flexibility where users can enable the latest features as they become available and never have to worry about adding hardware or upgrading their server which is all done automatically on their behalf as part of the cloud service.
Cloud solutions are not without their faults. Monthly costs can be high, especially if processor intensive services like video analytics or AI are required. Furthermore, they are critically reliant on the local cameras having constant access to a stable internet connection.
All three software locations have their merits, though none are perfect. In most cases, a blend of two of the three locations offers the best solution. Consumer cameras are best served with a blend of edge storage and cloud backups, as consumers are not interested in maintaining a server in their home. For commercial applications, the cloud is typically still too expensive where camera counts on a single site could number into the hundreds. It is unlikely any one solution will win in the end. For end users currently considering a new system, careful research needs to be put into calculating not only which solution is the best today, but which one will be able to evolve over time to keep up with new technology without costly changes to existing hardware
Colin Bodbyl is the chief technology officer of Stealth Monitoring.
This story appeared in the November/December 2018 issue of SP&T Magazine.