IP Video Design Considerations

Different requirements factor into bandwidth calculation

• By Mike Mathes
• Aug 01, 2011

When considering IP video design, a great deal of attention is usually given to bandwidth, processor throughput and storage.

Rightfully so, but many other implications of an IP video system are often overlooked.

When calculating bandwidth, processor throughput and storage, each camera, encoding algorithm and VMS has different requirements. It’s critical to the overall system that each of these issues is considered. But what makes an IP video system different from an analog system other than these obvious issues?

A primary difference versus traditional analog systems is the variety and flexibility of image resolutions and frame rates and how these options impact the overall system design.

Designers often use only resolution and frame rates to determine the bandwidth and storage requirements of the system. The implications go much deeper to the heart of the overall system design. Analog cameras provide less than 0.5 megapixels of image resolution. Today’s cameras provide 1, 2, 3 and 5 megapixels as a matter of course, with even higher resolutions available.

In an effort to shortcut the design process, we often see resolutions and frame rates that are much higher than necessary. This leads to increased system cost for servers and storage.

But it’s not just the first cost of the server. A server consuming 300 watts will cost more than $1,000 per year in energy just to run and cool it. Could the system be designed with fewer servers? With today’s green initiatives taking center stage, it behooves us to ask these questions and undertake an effective design.

So, what are the fundamentals of proper system design? It starts with understanding the application for any given scene. Based on the application, first determine the level of resolution needed. There are three general principles or levels: detect, classify and identify.

Detection. The system will capture general information. You’ll be able to see crowds, traffic flow or an individual object moving in the scene. General surveillance and detection requires a minimum of 6 pixels per foot.

Classification. If you need to be able to classify an object within a scene, the minimum pixel count is 20 per foot. At this resolution, you’ll be able to classify objects as people, animals and vehicles.

Identification. The highest level of resolution. Forty pixels per foot is considered the minimum to be able to identify an individual or read a license plate. Proper analog system design also should incorporate these principles. Internet Protocol, with its various resolutions, can provide greater flexibility, helping to reduce camera count and the associated infrastructure costs.

Once the resolution level is determined, you’ll need to calculate the total resolution required for a given scene.

Measure the width of the subject field at the required distance. Multiply the width by the pixels per foot required and you’ll have the number of pixels required to capture the scene. You can then determine what resolution the camera or cameras can provide to achieve the necessary coverage.

Another often-overlooked fundamental of the design is lens selection. Proper lens selection is critical, particularly when dealing with high-definition IP cameras. Lower-quality lenses will not focus to the resolution of a high-definition sensor. This reduces the effective pixel count, wasting money that’s been spent with the intent of achieving a higher resolution.

While it’s not the intent here to provide a tutorial on lens selection, it’s important to understand sensor size and format, and how that relates to an appropriate lens for a specific field of view and resolution.

With wider views come the need for greater depth of field, particularly in scenes where the view isn’t straight on. A road or path angling away from the camera presents the subjects at increasing distance as they move across the field. While this occurs with analog systems, a 5-megapixel camera can experience the effect fourfold and requires a more calculated approach when selecting a lens.

A higher pixel count isn’t always better. By following the above design principles, the pixel count can be optimized and lower-cost, lower-pixel count cameras can be used, keeping the bandwidth and storage requirements to a minimum. Remember that doubling the pixel count can double bandwidth and storage but will only increase resolution by 50 percent.

Another fundamental principle of IP design is frame rate. The tendency of designers and users is toward higher frame rates. Many cameras offer a host of frame rate adjustments right up to full-motion 30 fps. Because encoding algorithms MPEG and H.264 rely on scene differences, an increase in frame rate doesn’t proportionally increase bandwidth or storage. As the frame rate increases, the difference from one scene to the next reduces, and therefore less data is in the stream. There is a price to pay, however. Viewing workstations still need to fully recreate each frame. Higher frame rates and encoding such as H.264 put more demands on the processors in these workstations.

To help determine a desired frame rate, we can apply some logic. How fast is the subject moving through the scene? How wide is the scene? How many images of the subject moving through a scene do you need? One image of a person may not be enough for identification, but it could be for a license plate. Once you know these criteria, you can determine a minimum frame rate without overtaxing the system.

A final word on cost: many cost comparisons look at equivalent camera counts. By now you should realize that’s not an effective comparison. Costs for an IP system should be compared to an analog system of equivalent coverage. Often, with fewer cameras and reduced infrastructure, both the initial costs and ongoing operation of an IP system will be less than an analog system.

Today’s advanced IP technologies require an indepth understanding of system design. This goes much deeper than understanding the bits and bytes of bandwidth, processor throughput and storage. It is necessary to understand how to apply various camera resolutions and frame rates to design a quality video system.

Ensure your system designer and integrator have a solid understanding of these design principles before finalizing your next project.

This article originally appeared in the August 2011 issue of Security Today.