Bandwidth

Brian Karas
Vote 0 Votes
How much bandwidth do I need?
Can I use a DSL connection?
How much data do the cameras send?
Will this work with a 3G card?
How many cameras will a cable modem support?

These are all questions we hear frequently.  When dealing with IP devices - cameras, encoders, printers, whatever, it's understood that device is going to use some amount of network resources to communicate.  Cameras in particular have a reputation of being "bandwidth hogs".

Figuring out how much bandwidth is needed is not a complex calculation, but there is also no "one size fits all" answer.

VideoIQ devices send data across the network in three primary scenarios:
1) When an alarm event occurs - the alarm clip is sent to any PC(s) logged into that unit and/or to any FTP servers configured to receive push events on the camera

2) When a user is watching a live video stream from the device

3) When the Archive process is backing up recorded video.

A couple of terms that you often hear in relation to bandwidth calculation are "VBR" and "CBR".  

VBR is "Variable Bit Rate", meaning that the bandwidth used may change from one moment to the next.  Sometimes by a small amount, sometimes by a large amount.  VBR is common when you want to specify a fixed quality level and then let the camera use as much bandwidth as it needs to meet that target.  This sounds good on the surface, and there are good applications for VBR, but it also frequently results in cameras consuming massive amounts at unexpected times, and not always with obvious quality.  Changing resolution and/or framerate will tend to have a direct impact on the bandwidth used, since higher resolutions or more frames per second will require more data.

CBR is "Constant Bit Rate", the user defines the target bitrate for a given video stream configuration, and the streaming software compresses it using the selected format (H.264, MJPG, etc.) to fit in the allocated bandwidth.  Changing resolution settings for a CBR configuration will NOT impact the bandwidth used, since the bandwidth is configured to a fixed setting.  Image quality is what gets impacted.  With a CBR system, we can specify a target bitrate of say 2Mbps, which could produce a really nice image for a 720p image stream at 2.5fps, but would not be very good for a 1080p stream at 30fps.  Both image streams would use the same bandwidth (2Mbps), but the lower resolution stream would likely LOOK better, because it wouldn't be as heavily compressed as the 1080p stream. The benefit of CBR is that you get very predictable bandwidth (and storage) calculations, but need to understand what settings may or may not be optimal.

VideoIQ uses CBR bandwidth management on all current devices.  Bitrate settings for individual streams are configured in the Storage and Compression settings of the Camera Settings (accessed by right-clicking the camera icon in the tree.

BandwidthSettings.pngChanging resolution or framerate will NOT affect bandwidth or storage times.  However, selecting a high resolution and framerate in conjunction with a low bandwidth speed will result in image compression that may product noticeable degradation.  For 1080p streams it is recommended to choose a bitrate setting of 4.5Mbps or higher. 

There may be aspects of your network connectivity, especially over wireless networks, that prevent you from viewing a live stream above a few hundred kbps.  A setting of 500Kbps for the non-alarm stream is usually sufficient to produce a decent quality 960x540 pixel image at 2.5 or 5 fps.  Because of the way the VideoIQ devices operate, you can always choose to record the higher resolution stream, even if it won't be viewed live.  The Export feature makes it easy to copy out hi-res video in selected intervals.  Because this is a file transfer instead of a live stream at that point, you can export the videos even over a low speed connection, it just may take 3-10x as long as the export duration to transfer the video (eg: a 30 minute segment of video could take 2-5 hours to transfer over a low speed connection).




Recording Overview

The most common operational mode of the iCVR is to record video at a rate suitable to yield a usable image for basic viewing and verification, and then to do event-­‐based scaling to a higher resolution stream when an analytics rule is violated. We generally refer to the stream that represents the continuous recording rate as the "LowQ" stream, and the stream that represents the higher quality video as the "HiQ" stream.

The iCVR can also be configured to do continuous recording at the highest possible quality, or to initiate events, but not scale the video resolution up during these events. Although these are both perfectly valid use cases, they represent an atypical deployment, and will not be covered in detail in this document.

Because the iCVR's storage usage is dependant on the number and duration of events in a given time period, it can sometimes seem difficult to estimate the recording time you will achieve from a given camera model. In reality you can often calculate the expected storage time in a straight-­‐forward manner, this document is intended to assist in the understanding and calculation of iCVR storage times.

If you were to visualize the continuous LowQ recorded stream, it would look like the following diagram, essentially a video stream recorded at a near-­constant bitrate:


lowq1.png 

Now, if we factor in the analytics-­‐driven events, the recording timeline would be visualized like this:

hiqlowq.png 

The LowQ stream continues to record, and as events occur the iCVR records a secondary video stream (the HiQ stream) for the event duration. It starts to become apparent that total recording time becomes a function of the LowQ recording rate AND the number and duration of alarm events, along with the HiQ recording settings.

Understanding Recording Calculations

Because the LowQ recording is continuous, and generally the primary consumer of disk space, it's often helpful to first calculate the longest possible storage time, which would be a scenario where there were no alarm events (or where alarm events did not use the increased resolution stream).

The first rule of thumb is that every 100Kbps of recording rate uses ~1.1GB of hard drive space per day.

Note: The iCVR manages streams primarily by bitrate, not by resolution or framerate. You can of course configure different frame rates and resolution rates (depending on product model) for a target bit rate, but the bottom-­line factor is the bitrate.

It also stands to reason that a D1 5fps stream at 100Kbps would look better than a 960x540 15fps stream at 100Kbps, as the camera would be attempting to fit less information in the data "container" at the D1 (704x480px) resolution than at the 960x540px resolution. The iCVR attempts to maintain a near-­constant bitrate, however in cases of extremely static scenes the bitrate may dip BELOW the configured settings, and in extremely active scenes the bitrate may burst ABOVE the configured settings for brief durations. However, the average over time tends to be very close to the configured settings.

For any given hard drive in an iCVR camera or encoder, approximately 15GB is reserved for the video index database and other logging functions, the remainder being left for actual video recording.

Example: An iCVR SD Dome camera with a 160GB hard drive would have approximately 145GB for actual video recording. The LowQ stream on SD resolution iCVR products is 100Kbps, which equates to 1.1GB per day of recording space. Thus:

145GB / 1.1GB = 131.8 -­‐ We would get 131 Days, or a little over 4 months, of continuous recording if there were no alarm events.

The default alarm event recording rate for the SD units is 500Kbps. This is 5.5GB per day if recorded continuously, but alarm events are sporadic events throughout the day. 5.5GB/24Hours is .23GB, or 230MB per hour. Every hour of alarm events per day uses 230MB of space, in addition to the space the LowQ stream is using. One hour of alarm events per day would make the daily hard drive space consumed: 1.1GB + .23GB = 1.33GB.

145GB / 1.33GB = 109 -­‐ We would get more than 3 months of continuous recording if we averaged 1 hour of alarm events per day.

One hour of daily alarm events has reduced the total storage time by 22 days. As total alarm event time goes up, total retention time goes down proportionately.

Calculating Storage Time and Choosing Storage Settings

Various combinations of LowQ and HiQ recording settings will yield different amounts of hard drive consumption. The following charts show how much hard drive space would be consumed per day for .25,1,2, and 4 hours of alarm events per day. A site using the iCVR primarily as a video intrusion detection device will typically see very low amounts of alarm activity (even 15 minutes per day is a high average), conversely a site setup to create a higher volume of alarms may reach 1-­‐2 hours of average alarm events per day. 4 hours of alarm events per day is rarely encountered, and is illustrated here to give an idea of minimal storage times in different configurations.

To use these charts:
oFirst, reference the chart that most closely resembles your expected average daily

alarm rate

oSecond, find the column under the LowQ settings that correlates to the recording rate you will use for non-­‐alarm (LowQ) video

oThird, look across the HiQ row for the alarm recording rate you will use and find the intersection of LowQ column and HiQ row. This is the average daily hard drive consumption in gigabytes (GB) that you should anticipate.

oLast, divide the number from step 3 above by the hard drive size in your iCVR to get the anticipated days of storage.

Example: Using the 1 Hour chart with an iCVR-­‐HD camera. The LowQ recording rate is set for 750Kbps and the HiQ Rate is set for 3000Kbps, the intersection of these rates is 9.5GB. A 250GB iCVR-­‐HD would give ~22 days of storage (250GB HDD - 15GB reserved space = 235GB. 235GB/9.5GB per day = 24.7 days)

15 Minutes/Day Average Alarm Activity

15MinChart.png 

1 Hour/Day Average Alarm Activity

1HourChart.png    

2 Hours/Day Average Alarm Activity

2HourChart.png    

4 Hours/Day Average Alarm Activity

4HourChart.png    

You can also use these charts to find LowQ and HiQ recording options to ensure that you will select the appropriate hard drive size for a desired amount of storage time.

Example: You want 14 days of storage from an iCVR-­‐HD camera. You anticipate your environment will have about 1 hour/day of alarm activity. We can consider a 500GB unit has 485GB of available storage space, and a 250GB unit has 235GB of available storage space. 485GB/14Days = 34.6GB/day and 235GB/14Days = 16.8GB/day. Using this data, we would look for settings on the 1 hour chart that correlate to 34.6GB or 16.8GB (or less) per day. We can see there are several options that use 16.8GB or less, meaning we could easily expect 2 weeks of continuous recording time from a 250GB iCVR-­‐HD with most settings, even if some days the activity level is higher than anticipated.

Common Settings

The most common settings for the iCVR-­‐SD Cameras and Encoders is LowQ at 100Kbps (on the SD models, the LowQ stream is always 100Kbps) and the HiQ stream at 500Kbps (factory default). The 250, 500 and 750Kbps HiQ recording options on the SD cameras have been boxed off in the 100Kbps LowQ column for easy reference when looking up values for those models. For most applications, the difference in drive space consumption per day at 750Kbps vs. 500Kbps is negligible, so it may be worthwhile to bump up the settings to the higher rate to get more detailed video.

For the iCVR-­‐HD Cameras, the most common settings are LowQ at 750Kbps, and HiQ at 3000Kbps. Using the reference charts above, we can see that 10-­‐11GB/Day is a good estimate for iCVR-­‐HD drive space usage. 

Adding 1­‐way or 2­‐way audio to an iCVR or Rialto installation is a good way to increase the value and effectiveness of the security system. By installing speakers and microphones with the iCVR a remote operator can speak and/or listen to persons at a remote location. This capability can be used to better manage visitor access as a door intercom system, or can be used to communicate with and discourage unauthorized visitors from a monitored site.

The audio output of the is a 1VP‐P line­‐level (unamplified) signal. This is by far the most common audio output signal among IP cameras, and is compatible with a wide range of amplifiers or self-amplified speakers. For outdoor applications, typical amplifiers will be rated for 10­‐50Watts of output power. Power ratings and speaker type are dependent on the environment, other ambient noises in the area, and distance the audio signal needs to heard from.

The amplifier may be located near the speaker, near the camera/encoder, or in an indoor equipment room.  In many cases a self-amplified speaker is the most efficient approach. If several speakers are going to be connected to a single amplifier, it may be beneficial to research 25V or 70V audio systems, to minimize concerns of aggregate speaker impedance on the amplifier. If the amplifier or amplified speaker is going to be located more than 20' from the iCVR special care should be paid to route the audio signal wire from the iCVR to amplifier away from any sources of noise or electrical interference. Use of a properly shielded cable is also recommended.

For the audio input channel, the iCVR and Rialto devices expect a 1VP­‐P  line­‐level input. There are microphones available with direct line‐level outputs, or a microphone preamp can be used to provide the proper input signal level. Microphones will also typically offer a variety of different pickup patterns, and some also offer some background noise cancellation or other sound filtering features. Much like the audio output, selecting the proper microphone for a given scenario can be situation dependant. Factors such as background noise and distance between the microphone and the person or object the audio will be captured from will determine the best choice for the given jobsite. 

When the microphone is placed a long distance from the person speaking (such as on a rooftop or similar remote location) a more directional audio pickup pattern and higher gain microphone is generally used. Microphones placed at doorways or other confined areas will often use omnidirectional microphones. It is important that the installer test the proposed microphone in the installation environment, or a similar environment before proceeding with final specifications.  For customers using the audio-in features, it is recommended that "talkbox" style intercom stations are placed and clearly marked in the installation site.  Without advanced knowledge of microphones and sound system designs it is very difficult and time consuming to design a system that lets a remote operator listen-in on a large open outdoor environment.  

This technote is intended to describe basic audio components and connections. Many options for audio exist in the security industry, and Video IQ encourages the integrator to become familiar with the equipment before deploying in a production environment.

The iCVR Encoder and Rialto units utilize a single 3.5mm "stereo" connector for both the audio out and audio in connections. In order to access each signal individually you will need to connect a breakout cable to the iCVR. These cables are available online or from Video IQ. This cable connects to the Audio IO port on the Encoder or Rialto and provides a connection for the audio Input and Output signals.  The iCVR Encoder has a single Audio I/O connector, while the Rialto devices offer dual audio I/O's, meaning you can have 2 separate speakers and microphones, controlled independently.

The iCVR Dome Camera has the audio I/O on individual RCA jacks.  The color-coding matches the PC color code standards, green is audio out and pink is the microphone input.

Amplifier and speaker selection will often be dictated by to requirements of the environment. For door intercom type applications typical amplifier power ratings will be 5-10 Watts.  For outdoor loudspeaker cases you'll normally be in the 10-25 Watt range.  One of the most commonly used items for outdoor audio talk-down is the Valcom 1036-M self-amplified speaker.  This speaker has a built-in 15 Watt amplifier and is rated for marine environments, so it will stand up to most locations.  Be advised it uses a slightly funky -24VDC power supply, so be sure to buy the Valcom supply when you buy the speaker.  You can find them through several distributors, or online for around $100-$120 (not including power supply).  

vc-v-1030c_3.jpg


If you are using a separate speaker/amp combo, it's important to be aware of a few common speaker ratings. The most important criteria to pay attention to for speaker selection are power ­handling, generally measured in watts; and impedance, measured in ohms. The power-­handling capabilities of a speaker relate to how much input power the speaker can handle. Higher power/wattage ratings do NOT equate to louder speakers, it is simply a measurement of the maximum power a speaker can handle without being damaged. The speakers power handling generally does not have to exactly match the amplifier, a 15W speaker could be used on a 25W amplifier, as long as the amp was not turned up to maximum output power.

Impedance, or load, must be matched to the amplifier, 8 ohms is the most common speaker impedance in these applications. Using a speaker with a higher impedance will cause the audio volume to drop and may stress the amplifier. Speaker impedances that are lower than what the amplifier can support will often cause the power ratings of the amplifier to be exceeded, which can result in distorted or clipped audio, amplifier overheating or amplifier damage. 

For audio-­input from a microphone, it is often easiest to use a microphone with an on-­board preamp, or a matched preamp, that supplies a line-­level output. Louroe and Crown both produce microphones that fit these requirements. Louroe offers several options in their "Verifact" line of microphones, while Crown models with the -LL part number designation are line-­level output mics. 

Finally, you will need to enable the Audio Out and/or Audio In features on each iCVR that is connected to a speaker or microphone. This is done through the "Camera Settings" option, by right-­clicking on the camera's icon in View. Once in the Camera Settings window, click the "Audio" icon to set the audio properties for the device. You can use the sliders to adjust the output volume and input sensitivity for the iCVR. You can also test the audio output level by clicking the "Start" button, which will play a predefined audio clip from the iCVR out through the speaker. Click "Stop" when you have adjust the amplifier gain and/or audio output volume slider accordingly. 

ViewAudio.png