Bandwidth versus video quality; it’s the oldest battle in IP video system design. Calculating a system’s bandwidth load is a vital task, but it doesn’t have to be a daunting one. Answering the right questions is critical.
Question 1: What’s the Video System Use Classification?
Three primary classifications of IP video surveillance systems that directly affect bandwidth are observation, recognition and forensic review. What is the end-user hoping to accomplish with each camera in the system?
If the customer is installing/using an IP system for the first time, it’s the integrator’s responsibility to present a realistic depiction for 1. what classification is required from each camera and 2. what is possible with today’s technology within their budget.
Question 2: What are the Basic Video Quality Needs?
Once use classification is chosen for each camera, this will determine how you configure the three basic settings that balance bandwidth and user expectations:
- Resolution
- Frame rate
- Compression
When the user must review video data forensically, they will most likely benefit from the greatest amount of resolution. Observation, on the other hand, requires a lower resolution but with a higher frame (re-fresh) rate.
The more resolution and frame rate required for each camera, the more video data that traverses the network. This is where compression comes into play.
Video compression reduces and removes redundant video data so that a digital video file can be efficiently sent over a network. Video quality, however, can be affected depending on the compression technique and percentage—which is why you want to use the most efficient techniques available. H.264, for example, can reduce the size of a digital video file by more than 80 percent without compromising image quality compared to Motion JPEG and as much as 50 percent more than MPEG-4.
Determining adequate compression levels will rely on a test-and-see exercise to ensure the customer is happy with the video quality vs. the corresponding bandwidth levels.
Remember, if the end-user has a minimum spec for frame rate and resolution, the only adjustment you can make is compression. If it approaches the point where image quality is less than desirable, you’ll need to consider upgrading the network hardware.
Question 3: What’s the Scene Complexity?
Scene complexity is comprised of the activity and visual environment in the camera’s field of view, and it affects overall bandwidth on a camera-by-camera basis.
A camera aimed at a white wall will have far less image complexity than one overlooking a tree line on a fall New England day. This is why camera vendors try to select colorless and boring rugs for their tradeshow booths—since a patterned, colorful rug could spike bandwidth levels and make for a poor demo.
Next, work with the end-user to estimate how much activity they anticipate in each camera’s field of view. A camera at the front door will have much more activity than the one overlooking the loading dock.
Most manufacturers offer interactive system design tools to calculate bandwidth for each camera based on the video quality levels selected (frame rate, compression, resolution, etc.) compared against different scene scenarios (intersection vs. schoolyard vs. lobby, etc.).
Question 4: Which Cameras will Multi-stream?
The first three questions cover how the settings and performance parameters affect bandwidth levels for a camera’s stream. However, today’s IP cameras employ multi-stream capabilities. How many streams do you plan to use and where is each one going?
A basic use of multi-stream is to have one stream sent for live viewing while a second stream of the same scene is sent for storage. These streams are configured differently on a per-needs basis (i.e. the live view is typically lower resolution, while the storage stream is HDTV-quality for forensic video). This creates two separate streams on the network coming from one camera, and must be taken into account for calculations.
Question 5: How does the VMS Handle Multiple Streams?
Using event-based streaming is a way for the system designer to minimize bandwidth and storage. This is done by configuring the camera to only send a stream when detecting motion, audio, when a relay is triggered, etc.
When using event-based streaming, understand that video management software (VMS) providers handle multi-streaming differently. Certain VMS programs work with in-camera analytics to pull a single stream when an event is triggered. However, if the VMS is using server-based analytics, a separate stream is often pulled constantly so the video can be analyzed. A second recorded stream is then requested on the event. This is how the benefits of in-camera, or “edge,” analytics improves bandwidth efficiency.
A similar consideration must be taken into account when granting video access to different viewers. Higher end VMS providers can multi-cast video from the recording device to multiple different viewer clients (PC, laptop, iPad, etc.). One stream, multiple viewers. Lower end VMS programs, however, will call up a separate stream from the camera each time a viewer accesses the video.
Question 6: What Happens at Night?
Surveillance is a 24-hour a day need so it’s unfortunate how many people forget to test their installation at night. When light goes down, gain goes up. If not controlled, this can cause bandwidth spikes.
If the user requires nighttime surveillance in areas of limited lighting, the first step is to select a camera with advanced noise reduction algorithms (i.e. Lightfinder). After that, there are three considerations:
1.) Plan accordingly on the hardware side—most networks today are at least 1 GigE, which should be adequate. Bandwidth also equates to storage. Fortunately, storage today is much more inexpensive compared to five years ago.
2.) Adjust motion detection settings—what looks appears as video noise to us can be misinterpreted as motion by the camera. If the camera senses movement, it will encode the video and send to the recording device. This can cause 50 to 80 percent bandwidth spikes at night when there’s no real motion. Today’s best IP cameras employ advanced video motion detection to reduce these false alarms, and motion detection settings can be adjusted.
3.) Set ‘bookmarks’ for gain and shutter speed—tuning the gain and shutter speed settings can reduce noise. Set the camera’s min and max range to ensure that the bandwidth will not spike as the camera performs automatic gain control. Test the camera at night to find the right balance between motion blur and bandwidth.
Question 7: Will there be Advanced Uses?
How will the customer use the system? Will they use advanced analytics that require high resolution, such as license plate recognition? Can they employ cloud topology with supplemental edge storage to lower upstream bandwidth requirements? Are most of the users mobile (if so, see Table 1 for how setting resolution to mobile devices affects bitrate)?
A useful part of a more advanced video surveillance system design process is to identify bandwidth usage at various locations on the network. Command center, workstations and telecom workspaces might all have different video bandwidth consumption and varying network requirements. For such installations, an important design step is to plan for and verify all users have the video data they need, with room for expansion. A way to communicate this to the network infrastructure management team so quality of service (QoS) is maintained is shown below in a “Bandwidth Estimation Table” (Table 2).
Adding all the variables
Each security device will be asked to perform its own task. Each camera and VMS manufacturer handles bandwidth-related configuration settings differently. Each environment seen by each camera presents unique activity and lighting scenarios. Consider each variable by covering this checklist of questions and testing at each turn to accurately calculate bandwidth requirements.
