Introduction
One question in the minds of many users and potential users of video surveillance is, How do I calculate the quantum of data generated by a surveillance camera and what is the bandwidth required to stream live video stream to remote viewers?
Bandwidth Calculation
Calculating the approximate bandwidth of a video stream is a matter of multiplying the resolution of the captured image with the number of frames for the moving image. There are additional factors such as audio bandwidth and protocol (communication and compression) overheads; but, these do not add significantly to the bandwidth. Therefore, the applicable formulae are:
- Frame Size = Resolution * Colour Depth (Chrominance/Luminance information)
- Bit Rate = Frame Size * Frame Rate (fps)
If we take a camera capturing an image at D1 (DVD) resolution, viz. 720 x 480, and transmitting at 30 fps (frames per second); the relevant numbers add up as follows:
- Frame Size = (720*480)*24 = 829440 bits = ~ 830 Kbps (Kbits/sec)
- Bit Rate = 829440 *30 = 248832000 = ~ 250 Mbps (Mbits/sec)
This is the raw data bandwidth requirement. It is huge; and, it represents the bandwidth requirement of just one content generating source (a surveillance camera, in this case).
Video Compression
Along comes video compression, and the bandwidth requirement starts getting manageable. The popular video compression standards are MJPEG, MPEG4, and H.264 (also referred to as MPEG4-AVC). We will not go into the details of each standard, except to state that bandwidth reduction of anywhere between 80 – 99.5% is possible through the use of video compression.
All video compression standards work on the Constant Bit Rate (CBR) process, with minor variations. The level of compression possible is dependent on the video quality acceptable to the viewer; with a Bit-Rate figure encapsulating the bandwidth requirement for a given quality of video stream. The higher the Bit Rate, the better the quality of the video stream. The higher the compression, the greater the possibility of the presence of compression artefacts (frame break-up, blockiness, etc.) in the reconstituted image.
Let’s now apply video compression to the raw data bandwidth calculated above. The compressed bandwidth requirement, using the above mentioned video compression standards, looks something like the figures given in the table below.
| Video Compression Standard | MJPEG (Mbps) | MPEG4 (Mbps) | H.264 (Mbps) |
| Resolution/fps and video quality | |||
| 640*480/30 fps, High | 10.78 | 2.34 | 0.94 |
| 640*480/30 fps, Good | 8.91 | 1.88 | 0.7 |
| 640*480/30 fps, Average | 7.5 | 1.84 | 0.47 |
| 640*480/30 fps, Low | 5.86 | 1.17 | 0.47 |
Bit-Rates and Video Quality
Typically, video management applications allow the viewer to select the bit rate s/he would like to apply for the video stream, from the surveillance camera. InsightServer2, from Mistral Smartvue, offers the following Bit-Rate options:
| 64 Kbps | 2048 Kbps | ||
| 128 Kbps | 2560 Kbps | ||
| 256 Kbps | 3072 Kbps | ||
| 512 Kbps | 3584 Kbps | ||
| 1024 Kbps | 4096 Kbp | ||
| 1536 Kbps | 4608 Kbps |
Ideally, a remote viewer would like to settle for the highest bit-rate; however, we don’t live in an ideal world (unless the viewer has an 100 Mbps Ethernet pipeline to its location). A process of trial and error, though, soon establishes that the improvement in the quality of the video stream does not improve significantly (apparent to the human eye) above a certain bit-rate (in the case of the Mistral Smartvue cameras, 1.5 Mbps).So, if video stream quality is important, but bandwidth is at a premium, this is the bit-rate to go with. There is a caveat to this statement, though. If the quality of the connection, for a given bandwidth, is poor; the likelihood of dropped frames is high, and the playback frame rate may be lesser than the recorded frame rate (resulting in poor video quality).
Another way of working out which bit-rate to go with, is to take the bandwidth available, between the site where the cameras are located and the remote user, (let’s say 2 Mbps); and divide it by the number of surveillance cameras at the site (let’s say 7). Doing the math returns a figure of 292.571; in turn, pointing the viewer to a bit rate of 256 Kbps. The quality of the video stream may not be the best; but, the viewer gets a live video stream from all the cameras at the site under surveillance.
Conclusion
Remember, video stream quality and data bandwidth are directly correlated; but, a customer does not need to provision for balance-sheet breaking data pipes in order to view good resolution surveillance video streams.
