The Struggle Over VIDEO BANDWIDTH
Jul 1, 2005 12:00 PM, By Michael Fickes
Today, somewhere in corporate America, a CEO is dressing down a chief information officer (CIO) because it just took the CEO five minutes to send an e-mail message. The CIO blames the chief security officer (CSO), who has been connecting security cameras to the network.
Video technology is clearly migrating toward corporate networks. However, most companies use networks for vital operational functions, and CIOs tend to resist devices that consume bandwidth to the point of degrading performance. Important as it is, security video ranks among the most ravenous bandwidth consumers — just a handful of cameras can significantly slow the performance of many networks.
In recent years, security vendors have begun to solve the problem of bandwidth-hogging security video. Emerging solutions include DVRs that raise and lower the amount of video data sent over the network as needed. Other technologies compress video data into smaller files that are easier for the network to haul. Others take advantage of declining technology costs to build separate security networks. Camera makers are developing ways to manage video bandwidth requirements at the source — the camera. Still others are re-inventing how DVRs transmit video to the network.
“There are lots of bright people working on bandwidth problems, and there are many ways to address the issue,” says Peter Boriskin, director of technology for Tyco Safety Products, Access Control and Video Systems, Boston.
The bandwidth problem
The term bandwidth defines a network's data-moving capacity. Over the years, network capacities or bandwidths have risen from 10BaseT Ethernet to 100BaseT to 1,000BaseT. The term BaseT describes the twisted pair wire used to transmit data.
Today, most existing networks are 100BaseT Ethernet. This means they can transmit 100 megabits (Mb) of data per second. New networks generally provide 1 gigabit (1,000 Mb) of bandwidth, a significant increase over 100BaseT networks.
How many megabits does video consume? Video cameras take 30 pictures or frames per second (fps). When an encoder or a DVR converts each of those 30 separate pictures to digital data, the result is about one million 1's and 0's or 1 Mb. Twenty cameras then will require 20 Mbs per second of network bandwidth.
As a rule, CIOs want a company's basic operations to use no more than 60 percent of the total available bandwidth. The rest is reserved for surges, when everyone is online at the same time. For a 100BaseT network, then, 20 cameras will use up one-third of the capacity allotted to basic operations, which is too much, and the CIO will not allow it.
On a 1,000 Mb or 1-gigabit network, 20 cameras will use only about 2 percent of the 60 percent operating capacity. On the other hand, 200 cameras would raise security video bandwidth requirements back up to one-third, and the CIO would react.
Given how fast video can fill up a network pipe, it's no wonder the first question a CIO will ask about security video is: How much bandwidth will it use? Mitsubishi Digital Electronics America Inc. recently introduced the DX-TL5000U DVR that makes it possible for a security director to answer that question.
“The device contains a throttle,” says Jeff Kiuchi, a Mitsubishi Digital product specialist. “You can use the throttle to limit the amount of video data that will be transported. Speed settings include 20 Mb, 10 Mb, 7.5 Mb, 5 Mb, and so on, to as low as 100 Kb. This tells the IT manager exactly how much video each DVR will send over the network, making it easy to manage bandwidth usage.”
The Mitsubishi device can also reduce frame rates from 30 fps, while increasing video compression rates. To keep quality up, a software feature allows operators receiving the video at desktop computers to adjust picture quality.
Compression technologies tailored to security
Mitsubishi manages video transmission rates, while working a little with compression. Other solutions lower bandwidth use by redesigning the way video is compressed. Tyco's San Diego-based American Dynamics group, for example, recently introduced a bandwidth management solution called Intellex IP. This video storage and transmission device falls into a product category called network video recorders (NVRs).
Intellex comes with two network interfaces. One connects to a small network of up to 16 cameras. The other connects to the network. Like a DVR, the device receives video from cameras outside of the network. If the network breaks down, recording will continue. Unlike a DVR, which typically accepts signals from analog cameras, the Intellex only accepts digital video. To use Intellex with analog cameras, video must travel first through devices called encoders, which convert signals from analog to video.
In short, the Intellex technology creates an independent network for security cameras. Connect a dozen or so together, and the device can control a large network of cameras. “No video travels on the network until the Intellex sends it,” Boriskin says. “And when it does send video over the network, it uses a compression scheme called ACC.”
All digital video transmission systems compress video. The most common compression scheme is called MPEG, short for “motion pictures expert group.” Of several generations of MPEG, the most current is MPEG-4.
MPEG compresses video files by scanning each frame produced by a 30 fps video camera and tagging repetitive data. In a moving picture made up of 30 frames per second, each successive frame repeats a huge percentage of the preceding picture. MPEG systems transmit all the data from a few key frames along with instructions about how to recreate the 1's and 0's representing the video frames that do not change.
While MPEG compression eliminates tremendous amounts of data, there is still a large amount left. The Tyco-American Dynamics ACC Intellex replaces MPEG with a compression scheme designed for security video. “MPEG is good at compressing motion,” Boriskin says. “But it isn't as good on still photos. When you go to court, you usually show a still photo, not a video picture. ACC produces good still pictures, though you may lose some of the fluidity of moving images.”
In security, fluid moving pictures may not matter. Jerky video, for instance, can show a scene in which a man punches another man in the nose clearly enough for any viewer to understand the action. Then again, conventional security video often doesn't produce a photo clear enough to identify a suspect. ACC provides that level of quality, while compressing video into files almost 300 percent smaller than MPEG-4.
Managing bandwidth with cameras
“You can reduce bandwidth requirements by reducing the bandwidth consumed by the camera video stream itself,” says Steve Surfaro, group manager, enterprise sales with Panasonic Security Systems in Secaucus, N.J. “A new Panasonic IP camera connects directly to the network and manages bandwidth with two built-in compression technologies: MPEG-4 for low bandwidth, high frame rate use and mega-pixel JPEG for extremely high-resolution pictures.”
JPEG stands for Joint Photographic Experts Group and refers to a system that compresses still photos. The Tyco-American Dynamics compression scheme, ACC, is a variation of JPEG compression.
The new Panasonic camera goes a step further and records nearly 30 minutes of video on a removable media built into the camera itself. When the media fills up, it records over existing material — which has traveled over the network and been recorded by an NVR. The goal is to maintain video during network outages. When a network node goes out, the IT department will dispatch appropriate individuals to make repairs, Surfaro explains; meanwhile, the camera continues to record.
Building a separate security network
Some security needs require so many cameras that a CIO would never authorize mixing security video with regular network traffic. There are two solutions to this problem: a virtual local area network (VLAN) partitioned off from the main network, or an altogether separate network just for security.
Do security directors really want to rely on a network to manage security video? What happens when the network breaks down? “People ask about network failures all the time,” says Shahar Zeevi, director of product marketing for DVTel Inc., Paramus, N.J., which provides software solutions that manage security video. “When was the last time you had a problem with a network? A network is solid state, highly reliable, and redundant. The network is the most reliable part of the system today. Networks just do not crash anymore.”
Zeevi also says that declining technology costs have made it economical to consider building a separate security network.
Suppose, he says, that a company upgrades to a one-gigabit network. One gig offers a lot of bandwidth, and a CIO might be willing to carve out 200 or 400 Mbs for use by a security director with heavy video needs. “This would be partitioned off from the main network,” Zeevi says. “We've done this a number of times.”
If the existing network cannot provide a partition, Zeevi suggests building a network. “We always suggest working through the IT department on this,” he says. “That way, the IT department, which understands the technology, will manage the network. Sometimes, they can even share funds. This is true for partitioned or dedicated networks.”
Isn't it expensive to pull cable and purchase equipment to operate a network? Some systems require new cable, concedes Zeevi. But many companies have plenty of transmission infrastructure, some of which might be available to security.
“You can buy a 24-port 10/100/1000 — a one gigabit network switch for less than $500 today,” Zeevi says. “Suppose I connect eight encoders, each with eight camera ports to this 24-port switch. That would enable me to use 64 cameras. If I bought another network switch and connected another set of encoders with 64 cameras, my network would support 128 cameras. And the network switches would have cost around $1,000.”
Network equipment costs have fallen, agrees Surfaro, who goes on to say that it is more important to work with existing network specifications. “You may have to pay $3,000 for a switch,” he says. “You need a switch that can be programmed and maintained by your IT department and behaves like other switches they deal with.”
Reinventing the DVR
Whether the IT department allows security to use the existing network, creates a separate partition, or helps construct a new network, bandwidth problems will not go away, Surfaro continues. “What if you need 500 cameras?” he asks. “You can easily fill up any network with video.”
Conceding that point, CoVi Technologies Inc., Austin, Texas, recently introduced the Crystal HD, a high-definition video surveillance system that manages video off the network with a new DVR concept for uploading video to the network.
The Crystal HD system captures video off of the network in high definition. The system also provides outputs in a number of standard-definition digital resolutions, with lower resolutions requiring lower bandwidth for transmission over the network. The system will transmit video only when asked and only at resolutions requested by a system operator.
“Suppose you had a network of 500 cameras,” says Stuart English, CoVi's vice president of marketing. “Do you want to monitor all of them simultaneously? No. But you might want to monitor 16 at a time. Our cameras have high-definition sensors and provide outputs in high-definition and several standard-definition resolutions.”
The size of a video file rises and falls with the size of the frame on a video monitor. Crystal can send 16 camera views to be displayed on a single monitor in small frames. Each of the 16 frames uses 80 Kbs of bandwidth, for a total of 1.28 Mbs.
“If you develop a keener interest in four of those pictures, you can request that and still use only about 10 Mbs of bandwidth,” English says.
The system allows a user to scale resolutions up to view single cameras and down to look at multiple cameras. The system decides based on the number of camera views requested by the user. “At the end of the day, you can't increase video resolution unless you solve the bandwidth problem,” English says. “Our system re-invents the DVR structure in a way that does this.”
ABOUT THE COMPANIES
For information, circle the Reader Service number (listed below) or visit securitysolutions.com
|CoVi Technologies Inc.||45|
|Mitsubishi Digital Electronics||47|
|Panasonic Security Systems||48|
|Tyco Safety Products and Systems||49|
Want to use this article? Click here for options!
© 2014 Penton Media Inc.
Today's New Product
In support of the Privaris family of personal identity verification tokens for secure physical and IT access, an updated version of its plusID Manager Version 2.0 software extends the capabilities and convenience to administer and enroll biometric tokens. The software offers multi-client support, import and export functionality, more extensive reporting features and a key server for a more convenient method of securing tokens to the issuing organization.