Advantage: IP

• By Steven Titch
• Jan 07, 2008

IP-networked CCTV cameras have become more cost effective than analog in installations requiring as few as 40 cameras, according to a report released in September.

The study, which created a mock request for proposal (RFP) covering a 40-camera installation in a school environment, found that the IP system’s total cost of ownership (TCO) came in 3.4 percent lower than an analog-based solution.

The study, funded by Axis Communications, a manufacturer of IP cameras, but conducted independently, attempts to show the overall cost savings and benefits that derive from digital network centric deployments, despite the fact IP cameras themselves are more expensive per-unit than analog counterparts.

The study also underlines the dramatic decline in the cost of IP cameras to the point where many of the benefits that accompany digital video, ranging from plug-and-play components to basic analytics, are within reach of operations that require less than 50 cameras.

“TCO for installations above 32 cameras is 5 percent less than analog," says Fredrik Nilsson, general manager for Axis. “Between 16 and 32 is a wash." For one to six cameras, Nilsson adds, IP is still about 10 percent more.

One exception, however, is that if IP infrastructure -- such as pre-installed Category 5 cabling -- already exists at the customer location, the cost of an IP system will always be cheaper, the report found.

Offsetting Camera Costs

The cost of IP cameras is only part of the equation. While network cameras are 50 percent more expensive than analog cameras, cabling is almost three times as expensive in an analog system (see graph), chiefly because analog cameras require separate cables for power and PTZ control. Furthermore, the total cost of installation, configuration and training is almost 50 percent higher in the analog system.

In addition to measurable cost benefits, Nilsson cites “soft" benefits that increase the value and ultimate return on investment network-based systems offer. These include superior scalability, flexibility in placing and moving cameras (all they need is an Ethernet plug or failing that, a wireless local area network connection) and image quality. IP cameras also can be remotely serviced, and management systems are based on PC servers from firms such as IBM, Sun Microsystems and Hewlett- Packard that usually have superior warranty and service plans compared to DVRs.

Bill Stuntz, vice president and general manager of Cisco System’s Physical Security Business Unit, agrees with the TCO equation.“ You’ve got to look at the price of [IP] cameras in terms of the overall system," he says. “While it’s true that the price of the camera has been the hold-up, quality and compatibility is where you see them as the better option." Plus, he adds, one off-the shelf server can support many more digital cameras than can a single DVR, lowering the cost of ownership. “You also stay current through upgrades, versus replacement."

Entry Level Analytics

Another key “soft" benefit of network cameras is that they can store and process information like any other IP client, such as a PC or handheld PDA. Combined with the declining cost of ownership in general, the potential of the IP cameras to work as edge network devices has touched off a flurry of  “entry level" video analytics that vendors are pricing within reach of smaller users.

The trend gave rise to the buzzword, “generic analytics" on the floor of the 2007 ASIS International Seminar and Exhibits in September, which, in addition to Axis and Cisco, drew comments from companies such as ioimage and IQinvision.

Generic analytics, says Nilsson, is basic low-end functionality. It takes in people counting, “virtual fence" applications and tamper resistance. Axis itself introduced a tamper-resistant feature on its camera line at ASIS designed to alert monitoring systems if a camera’s field of view is changed, or if its lens is covered, obstructed, spray painted or vandalized.

“Analytics is still considered complex. People are a bit afraid of it," says Dvir Doron, vice president of marketing at ioimage, which has just introduced the iobox trk1, a single channel MPEG4 encoder that supports analog and digital cameras, to the U.S. market. “We pre-wrap everything in the encoder and camera and make it easy to install and obtain information."

The trk1 is designed for plug-and-play set-up and operation and is addressable with a standard Web browser. The device provides a wide-area solution that is uniquely accessible to small sites, as well as for supplementing larger installations. Analytics capabilities are largely trip wire and virtual fence functions, says Doron, such as intrusion detection, entry detection and sterile area protection.

“There is a need for a more affordable product to cover the one-channel world, versus four and eight channels," says Doron.

However, Paul Bodell, chief marketing officer for IQinvision, another manufacturer of IP cameras, advises users not to get hung up on price points. Deriding the notion of claims of “30 frames per second, 30 days of storage at $30 a camera," Bodell says the issue is how technology can solve a customer’s problem in a cost-effective way. “Price points have to be in line with the problem you’re solving," he says. “Expectations need to be in line with real-life parameters."

When it comes to analytics, he says, users already expect too much. For purposes of video and analytic processing, Bodell measures resolution in terms of pixels per foot. For general surveillance, including people and vehicle counting, users can get by with 10 to 20 pixels per foot. For forensic detail, such as face and license plate recognition, cameras need to be optimized to 40 megapixels per foot. Casino applications where chip and bill denominations need to be discerned require even finer resolution. Resolution and processing this fine remains expensive.

Axis’ Nilsson is quick to agree on managing expectations, but suggests that the perfect not be made the enemy of the good. The quality of video analytics, he says, is largely a function of processing power, which, according to Moore’s Law, doubles every 18 months. “Ninety percent is better than nothing," he says when it comes to analytic performance. “For some that’s quite good. For others, their requirements might take a few more years."