Fredrik Nilsson is general manager of Axis Communications and an authority on IP surveillance.
[Editor's Note: The following is the fourth in a series of articles looking at the "myths" that surround the world of IP surveillance. Fredrik Nilsson, the general manager at IP surveillance manufacturer Axis Communications, has authored 10 articles dispelling these myths. In 2005, look for them monthly in SecurityInfoWatch.com and in the Security Frontline newsletter, and in the months of March, June and September in Security Technology & Design magazine.]
IP-based surveillance is rapidly replacing and upgrading traditional analog systems. Industry analyst, J.P. Freeman and Co., estimates that network camera sales will exceed those of analog cameras by 2007 and that network camera sales will more than double those of analog camera sales in the network video market by 2008.
When evaluating network video technology, one of the most common arguments against it is that transferring video over a network will overload the network, causing problems with other mission critical applications on the network. There is, in fact, some truth behind this myth. Video can consume large amounts of bandwidth, affecting the overall performance of network. However, with any size network video installation, users can take a few simple steps when designing their systems to ensure that their IP-based surveillance systems will not overload their networks.
For instance, how much bandwidth a network camera will use depends on several factors, including image size, compression, frame rate (images per second) and resolution. Network video products will utilize bandwidth based on how they are configured. Looking at resolution, a high-resolution picture (4CIF) contains four times as much data as a normal picture (CIF). A reduction of the frame rate to half (for example, 30 frames per second down to 15) will reduce the amount of data by half as well. Additionally, because of built-in intelligence, many network cameras will only send video over the network if the video is worth recording, which might only be 10 percent of the time. Ninety percent of the time nothing is being transferred over the network.
Most small-scale installations - with only a few network cameras or video servers - can operate over an existing network. Because video runs on the same network as all other data transmissions, users should configure their cameras so that high-resolution images are not running at 30 frames a second over the network at all times. This would unnecessarily use up bandwidth and slow down other applications.
A professional network camera can send up to eight Mbps (megabits per second) of data over the network, depending on compression, size and frame rate. In order to reduce this, users can utilize the built-in intelligence in the network camera to reduce the size and speed of images transmitted over the network. A network camera can be configured to make "decisions" about video resolution and frame rate, depending on factors such as motion detection and time of day. For example, motion detected at 1 a.m. on Saturday morning - when no one should be in the office - can trigger the camera to transmit the highest resolution video at the highest frame rate. On the other hand, motion detected at 1 p.m. on Tuesday afternoon, would be considered "normal" and would not trigger an increase in resolution or frame rate.
Today, there are many examples of successful network video installations with hundreds, or even thousands, of cameras. For example, departments of transportation in Minnesota and New York use video servers to digitize the feeds from hundreds of analog cameras, enabling authorities to monitor roadways and commuters to view traffic conditions. School districts are also known for large-scale network video deployments. Districts often have several campuses that are spread out over large areas, which makes network video an ideal security and surveillance solution. Often times, schools piggyback network video onto underutilized data networks or even voice over IP (VoIP) systems.
Unlike small installations, enterprise-level deployments cannot always plug directly into an existing network. These extensive installations require that users take additional steps to ensure that IP surveillance technology will not tax the network.
If the decision is made to use the existing network, then it is best to define the minimum and maximum bandwidth available for the network video system. Enterprise networks consist of multiple segments of different speeds; a connection to a switch may be anywhere from 10 to 100 Mbps, while the backbone communicating between the two switches may range from 1 to 10 Gbps (gigabits per second). A 1 GBps network can transmit video from hundreds of network cameras, even at full frame rate, over a single network connection.
If the network video system is large enough, a separate network to handle the video transmissions will be required. This is similar to rail transportation: once the existing track becomes too congested, you simply must build another set of tracks. However, because network video operates with standard networking equipment such as switches and routers, separating networks is typically an easy and inexpensive process.
In addition to mapping out potential bottlenecks or building a separate network, enterprise users can rely on some of the following methods to better manage bandwidth consumption:
Switched networks: If many devices are connected to the same network, the network should be divided into segments with switches or routers placed in between. Switches sometimes have built-in router functions. Network switching - a common networking technique - separates one network into two autonomous networks. Even though these networks remain physically connected, the network switch divides them into two virtual and independent networks: one for data and one for video. By designing the system wisely and splitting the number of cameras between different sections or links, the user gains the benefits of higher reliability and improved performance.
Efficient compression: At high continuous frame rates, above 15 fps, considerable bandwidth savings can be achieved by using MPEG-4 compression, rather than Motion JPEG. The two formats usually target different applications, and MPEG-4 is not expected to replace Motion JPEG. However, MPEG-4 is recommended for live viewing and for applications where bandwidth and storage limitations are important factors.
Faster networks: As the price of networking equipment continues to fall, Gigabyte networks become increasingly affordable. Having a faster network alleviates bandwidth concerns, and a faster network increases the value of running security and surveillance applications over networks.
Event-driven frame rate: Full frame rate, 30 frames per seconds (fps), on all cameras at all times is more than enough for most security and surveillance applications. With the configuration capabilities and built-in intelligence of network cameras and video servers, frame rates under normal conditions can be set lower, at approximately 1-3 fps, to dramatically decrease bandwidth consumption. In the event of an alarm, the recorded frame rate speed can be automatically increased to a higher frame level.
Ultimately, a network's ability to handle the demands of a network video system depends on its configuration. It is important to take the time to consider how your network will operate when video is added and ensure that you are properly equipped to handle the extra bandwidth requirements. Although it seems intimidating at first, bandwidth issues can be easily be overcome with a little bit of advanced planning and proper configuration.
About the author: As the general manager for Axis Communications, Fredrik Nilsson oversees the company's operations in North America. He can be reached at firstname.lastname@example.org.