Simplifying the Complexity of IP Video

[IMAGE]Ss12ip1.jpg[/IMAGE]

In the days of analog-only video surveillance installations, things were simpler. A common piece of coax cable connected every camera to a monitor and each component in the system had standard BNC connectors. Everything in the system had its place and interaction between components was limited to its connections. Most significantly, everything was consistent from system to system. 

With IP video surveillance systems now running on networks, everything has changed. The IP system paradigm stands in complete opposition to its analog counterpart. Every part of an IP system interacts with all system components, which affects quality and performance.

We will take a look at some key areas of consideration for IP systems, as well as their affect on systems design. Being able to determine the actual amount of usable bandwidth will help minimize hang-ups down the road. By identifying power over Ethernet (PoE) Plus standards, operating components and video management software, you will be able to understand network expectations and capabilities.

Determining Usable Bandwidth

IP systems are based on networks, and bandwidth is the key component of any network system. In general, the more bandwidth there is, the better the system will perform. Often when questioning how much network bandwidth is available, do not just look at the equipment’s network connection specification. Don’t assume that 100Mbps or 1,000Mpbs (1Gbps) means just that – the reality is very different.

Security systems are continuously uploading and downloading information. Downloading usually takes the form of video transmitted from a source, and uploading commands from a client to the source. As an IP camera downloads its video information to a server, that server receives the video and uploads instructions to download to the camera. The server then downloads video to a client, which in turn uploads instructions to the server as to which videos to download.

In general, the process of downloading absorbs about 30 percent of network bandwidth in overheads, and uploading takes an additional 10 percent. As a result, a 100Mbps network, for example, would be left with about 60Mbps of actual bandwidth. About 10 percent should be allocated to system variations for differences in wiring and connectors. The result being that the 100Mbps network only has about 54Mbps of actual usable bandwidth. A 1Gbps connection would be left with about 540Mbps.

At D1 resolutions (720 X 480 for NTSC cameras), a good quality signal will usually require a bandwidth of about 1.5Mbps. The user should not make the mistake of relying on the 100Mbps figure, but focus on the more accurate 54Mbps estimate instead. That works only if the front end (the connection from cameras to server) and back end (the connection from server to clients) are on separate networks.

Another potential bandwidth mistake is assuming that both front and back ends are on the same network. Many software management systems offer the ability to view multiple cameras on a single screen. Some even offer the ability for an individual client to have four monitors.
The required client bandwidth is a function of the resolution of each camera viewed. The misconception is that viewing a D1 resolution image in one-quarter or one-eighth screen view takes up less bandwidth. However, signal reaching the client is still D1 resolution and its quality is based on bandwidth. A good-quality D1 image still requires about 1.5Mbps. Four images in a quad screen require 6Mbps; an eight view is 12Mbps. If the system has two monitors, double that requirement. If there are two clients with the same viewing capacity, double the bandwidth requirements again. 

Running Out of Bandwidth

What happens when there is not enough bandwidth? With some software management, there is a choice between maintaining the number of images per second and maintaining picture quality. Some systems will automatically reduce the image-per-second rate in order to maintain quality, which is often the best approach with video surveillance when optimizing the ability to identify individuals, license plates, etc. This is the reason why, regardless of settings for recording images, a system will provide the maximum number based on bandwidth and not software settings.

The situation becomes more acute when megapixel cameras are taken into consideration. A 3-megapixel camera at 30 images per second can require a bandwidth of up to 30Mbps. Apply this to the available 54Mbps of a 100Mbps network connection and you can easily see the problem.

Megapixel cameras provide high resolution with high quality at the expense of bandwidth. To compensate for this, you once again have to favor images per second for quality. In reality, if you are considering megapixel cameras, you should be thinking about a 1Gbps network. Don’t assume you have total access to all the bandwidth specified at the port. Even if you have 100Mbps or 1Gbps, you may not have it all to yourself.