Along with the potential impact on the information security of a client’s network, the bandwidth “bugaboo” is the most frequently sited objection raised by IT personnel when considering the connection of IP video encoders and cameras. To properly understand how to address this concern, it’s important to know exactly what amounts of network bandwidth are required by devices using the typical video compression programs available from mainstream surveillance camera/encoder manufacturers.

All video that is transported through a network must be “compressed” using one of several available compression formats. If NTSC video were to be transported through a network without compression it would require more than 150 Mbps of bandwidth, which is more than standard 100 Mbps Ethernet devices and switches can handle. The compression process is highly complex; however, the net result is that individual images are processed, redundant information is eliminated (or reduced), the number of individual pixels sampled can be changed in the compression settings, and the amount of compression applied to each pixel sample can be increased or reduced. Along with manipulation of the compression settings, lessening the frames per second being transmitted will result in lower bandwidth usage. Detailed below are the two most popular types of video compression utilized by IP security cameras and encoders. Whether a device is a complete IP camera or an encoder to which analog cameras can be connected the compression and bandwidth issues are identical.

JPEG (or MJPEG)

Common in most IP-enabled security cameras is the Joint Picture Experts Group or JPEG compression codec. This format compresses each image individually and completely, with “whole” pictures being transmitted over the network.

Here is a bandwidth measurement from a typical JPEG network camera, in this case an Axis 207W Wi-Fi camera:

When studying these graphics please note that the bandwidth scale is measured in bytes, not bits. One byte equals eight bits. Therefore, in the above image, the camera is putting out approximately a ½ megabyte stream, which calculates into a 4- megabit bandwidth usage.

MPEG-4

Motion Picture Experts Group (MPEG) is a very popular family of compression formats, used in DVDs, camcorders, and high-end IP cameras and encoders. The key to the MPEG formats is that only a small percentage of the images being transmitted are actual whole pictures; most are partial images, which only include any changes from a previously transmitted whole image.

Most manufacturers are providing MPEG-4 or its new variant H.264, which is also called MPEG-4 AVC (Advanced Video Coding) in their high-end products. This format has many strengths including the ability to adjust the transmitted frame rate from the NTSC 30 frames per second standard to a lower value. Given a reasonable quality analog camera with decent lighting, usable images can be transmitted using as little as 50 kbps of total bandwidth, while high-quality images can be achieved at bandwidth rates of 500 kbps or more.

The bandwidth output of an MPEG-4 encoder (Bosch VIP X1) at 15 frames per second looks like this:

When compared to the JPEG graph, the reduced bandwidth requirements of MPEG-4 are readily discerned.

Megapixel Cameras

While standard image (640 x 480 JPEG or SIF MPEG-4) cameras and encoders can require reasonably small bandwidth on a network, the megapixel camera is another beast entirely. Most megapixel cameras use JPEG compression to provide the highest quality images for viewing and recording. However, the cost is in bandwidth consumption.

Here’s the bandwidth output of an Axis 223M megapixel camera:

So the pretty pictures come with a hefty bandwidth price tag.

Bandwidth busters?

As with many aspects of life, there are myths and scientific realities that often don’t coincide with each other. The myth is that network cameras and encoders will overtax the capacity of a network, bringing down the system or dramatically slowing communications. The reality is that modern enterprise networks have much more overall bandwidth capacity than is needed to transport both multiple IP video streams and enterprise data traffic.

Consider the example of a single IP video encoder using MPEG-4 compression and transmitting a robust 2 Mbps video stream. It is a 100 Mbps Ethernet device, and is connected to a network switch. The first leg of communication, from the encoder to the switch is capable of 100 Mbps bandwidth; only 2 percent of that capacity is used to send the video signal to the switch. In a modern network Ethernet switches are generally connected using gigabit Ethernet trunks, so if the video stream passes from switch to switch it only uses two-tenths of one percent of the available switch-to-switch bandwidth.

If 20 encoders of the type described were connected to the same switch, the math stays the same, except now the encoder streams will be using 4 percent of the trunk capacity between the switches.

The real issue isn’t the bandwidth between the switches or devices; there’s plenty available. What can gum up the works is an individual switch’s capacity to process data packets. Every switch has a rated “throughput” which quantifies how much data that model of switch can process in one second. Network IT personnel can determine the effect of large amounts of surveillance video packets by confirming their switches throughput capacity and their current utilization by enterprise data traffic, and calculating the amount of video packets that will need to be simultaneously processed through the network.

It is important to understand that LANs often have large bandwidth capabilities, while Internet connections are comparatively much smaller. Take a look at this typical enterprise network schematic. While the desktop computers are connected via 10/100 Mbps Ethernet, and the backbone between the switches is gigabit (1000 Mbps), the T-1 Internet connection provides a comparatively lowly 1.54 Mbps upstream and downstream bandwidth. So if the LAN network provides a “superhighway” of bandwidth, the Internet connection is like a dirt road. This fact of Internet bandwidth means that for IP security video systems transmitting video over the Internet, a realistic value is three to five frames per second.

Video Bandwidth Management Options

There are a number of settings available on typical IP video encoders that can reduce the bandwidth of the related video streams. Let’s look at some of the options, and their potential effects on bandwidth usage and video recording/viewing:

1. Cut the Frame Rate

Reducing the frames per second rate from 30 to a lower number will dramatically reduce the bandwidth needed for a particular video signal stream. Enterprise networks will have a practical limit on how much data can pass through them, and frame rates lower than 30 fps are often a necessity. As a rule of thumb, typical network video encoders (and IP cameras) are often set at 15 fps, which still provides a high-quality viewing experience.

2. Jack up the Compression Ratio

Different devices will call the settings by different names, but most IP-enabled video encoders have some selection for how much the video is compressed. Higher compression levels lower the bandwidth needed while sacrificing image quality. A slight increase in the rate of compression may not be noticeable, particularly if the client is viewing the image on a desktop computer screen. Very high compression rates will result in images with “artifacts” — square block-looking sections of the image where the compression process has not completely processed that piece of the image. Security dealers need to be careful with these settings, as bad video may be worse than no video.

3. Edge Recording Option

Some devices have the capability of a local connection to a hard drive, typically via a USB port. These devices can be programmed to send a video stream to the locally connected hard drive for 24/7 storage, which then can be transferred into a larger NAS (network attached storage) array. This transfer usually can be timed to occur during off-hours, so the passing of a large amount of compressed video data does not impede enterprise data throughput. Most such devices provide some type of override so that if triggered by alarm inputs, video analytics, or an operator, the video stream becomes live.

4. Video Analytics

One savior for this video bandwidth concern is the implementation of video analytics into IP cameras and video encoders. Encoders can be programmed with analytics so that if there is no actionable activity being viewed by the camera, it transmits nothing over the network. As technology progresses, analytics will become cheaper/faster/better and will likely greatly reduce or eliminate network video bandwidth concerns by 2015.

Know your Options

One popular option in high-end devices is multiple video streams, where perhaps a low-bandwidth feed of 5 fps is transmitted over the network to be stored, while a high-bandwidth 15 fps stream is available for occasional live viewing by authorized users.

Manufacturers are producing high-quality IP video products with a wide variety of compression formats, settings and video stream possibilities. Understanding how to lower the bandwidth requirements while maintaining a usable viewed/stored image is a vital key to the success of a network-centric surveillance video installation.

Sidebar: “This doesn’t look like Terminator 2!”

It is very important when managing IP video bandwidth to demonstrate to the buyer how the images will appear on their local computer/monitor and over the Internet. Many clients, accustomed to having high definition TV, may mistakenly believe that purchasing a networked surveillance video system will give them the same viewing experience they receive as when they’re watching television at home. This is not the case, and astute security integration professionals will provide a live demonstration that will manage a client’s expectations.