Transmitting high-def video on low-bandwidth networks
Editor / Provider: Submitted by IndigoVision and Vemotion | Updated: 10/9/2012 | Article type: Tech Corner
As IP cameras increase in resolution and customers demand more clarity, network bandwidth becomes a critical issue. IndigoVision and Vemotion Interactive discuss how careful bandwidth management is needed to get the best out of your IP video network.
HD cameras now produce more data than a wide-area network (WAN) can reasonably handle. While your in-house local-area network (LAN) may be able to handle large amounts of data, it is important to be aware of the limitations when you want to send camera streams to other buildings or companies; the bandwidth available will be lower, and you need to be clever about what you send. With the growth in sales of ever-better HD cameras, there is a need for even better bandwidth management to make the best use of resources on any multisite system.
Avoiding bandwidth bottlenecks involves looking at five key areas:
- Video compression
- Choosing what to send
- Location of recording
- Managing demand
There is a lot of misunderstanding in the marketplace about video compression. Use of the H.264 video codec standard is usually seen as the answer, but H.264 is very broad term. It really only describes the format the video is compressed to, and how it can be decoded; it does not give information about how the compression was achieved. There is a big difference between good compression and bad; the table below details typical data rates for a single 1-megapixel camera monitoring a fairly static scene such as a building entrance:
It is important to do your homework and benchmark the compression from different suppliers so that you really know what you are buying. Great compression uses up to half the bandwidth of good compression. As well as reducing overall bandwidth, good compression also allows you to dual stream – high frame rate and high bandwidth locally, and low bandwidth for what you stream over the WAN.
An apt example would be the Vemotion Interactive VB-40 series that requires a GPRS, 3-G, ADSL/Ethernet or satellite router. Inputs and outputs (16 maximum outputs per unit) are configured and transmitted over the network to the control hub. The unit will connect to the server application and then, by installing a multiscreen viewer application on a PC or mobile device, you can view multiple live video feeds from wherever you are. Full telemetry for control of cameras is provided, and automatic bandwidth adaptation offers maximum flexibility.
For example, 16 simultaneous streams can be viewed from each VB-40 unit, and the operator can choose to increase the resolution of any of the feeds to provide enhanced video quality. The VB-40 will respond by automatically changing the video quality of other feeds in order to give the required channel the most bandwidth. This means that a fixed bandwidth can be intelligently distributed across all the video feeds in order to provide the optimal operating solution for the user. VB-40 can be “dropped” into location and give immediate access to camera systems already in place — from a private surveillance hub for temporary remote access, or to connect multiple locations and control rooms together where other network options would be cost-prohibitive.
Latency is another vital consideration. In recorded video, it does not matter if there is a little latency in playback of video from recorder to viewing station, but if an operator is trying to track real-time action and there is a double delay – delay in the images getting to the monitor, and then delay in the signal from the operator's joystick to the PTZ camera following the events – he or she is going to quickly lose track of what is happening. The time from the moment the light hits the camera lens until the image reaches the operator's monitor should be less than 100 milliseconds. Longer than that, and the lag is too much.
This is, in fact, one of the reasons analog systems are still around in so many areas – latency can be a real issue with IP video. Only a really well-designed system, carefully pipelining every stage of compression and transmission, will minimize latency properly.
Choose What to Send
Only send it if you have to: do not overload your network resource with data you do not really need. A good IP security management system will switch the transmission link into "idle mode" when there is nothing happening in the scene. The instant anything happens, things should switch back to full speed, and your system must make sure that is a fast changeover so that nothing is missed in the interim.
Real-time video analytics are vital too; they can identify when something relevant happens, perhaps when someone walks the wrong way down a corridor, or an item is moved. That is when you want to be sure you see everything that is going on.
Location of Recording
Recording location is another important consideration. Do not move all of your data back and forth between cameras and a central site. Design your network with NVRs close to the cameras they are recording, using your plentiful LAN bandwidth, and only transmit important information over the WAN.
Out of any video you record, you are typically only ever going to look at less than 1 percent, so keep the intelligence at the edge and only access what you really need to. A distributed IP video architecture, without the need for central servers, lets you record locally but view from anywhere you need to. Careful design will eliminate a lot of bottlenecks and headaches.
This is possibly the trickiest part to get right, and the one that most people forget about: managing operator demand. This is crucial on any multisite system. Each operator has to learn how to share the bandwidth, and to make sure they get their job done without hogging limited resources. Depending on the product used, this can often be managed within the management system itself; a built-in understanding of what each area needs can be configured during setup and managed silently so that bandwidth is allocated well without operator input. However, staff must still understand the need to share and may have to be encouraged to do so.
On top of all these recommendations, design your monitoring solution carefully and think about what is really important. There has been a move recently toward using HD cameras just because they are availabe and getting cheaper. But once you start putting 5-, 10- or 16-megapixel cameras in every situation, you are going to use up valuable bandwidth very quickly. It is worth bearing in mind that many higher-megapixel cameras are designed to work at slower frame rates; a lower pixel count but higher frame rate may actually be better at gathering evidence.
It is all about balance: choose the right mix of SD and HD cameras for your real needs, put them in the right place, and design the network to work well with them.