Monday, August 25, 2008

Bandwidth Basics for IP CCTV Design

When using IP cameras, Megapixel cameras, NVRs or even DVRs, understanding the basics about how much bandwidth is available and how much is needed is critical in planning, designing and deploying IP video surveillance systems. Everyone in the industry should have an understanding of the basics as bandwidth is a critical factor in video surveillance
This article is focused for a non-IT audience such as security managers, electronic technicians, sales and marketing folks. I am purposely ignoring details and edge cases to help a broader audience better understand the basics.

How Much Bandwidth is Available?
To figure out how much bandwidth is available, you first need to determine what locations you are communicating between. Much like driving, you will have a starting point and destination. For example, from your branch office to your headquarters. However, unlike driving, the amount of bandwidth available can range dramatically depending on where you are going.
The most important factor in determining how much bandwidth is available is whether or not you need connectivity between two different buildings. For instance:
In the Same Building: 70Mb/s to 700 Mb/s of bandwidth is generally available
In Different Buildings: .5 Mb/s to 5 Mb/s of bandwidth is generally available
The amount of bandwidth available going from your office to a co-worker's office in the same building can be 200 times more than the bandwidth from your office to a branch office down the block.
This is true in 90% or more cases. Note the following exceptions:
If these are different buildings but on the same campus, more bandwidth may be available.
If you are in a central business district of a major city, more bandwidth may be available.
If you are a telecommunications or research company, more bandwidth may be available.

Different Buildings
The key driver in bandwidth availability is the cost increase of deploying networks between buildings. Generally referred to as the Wide Area Network or WAN, this type of bandwidth is usually provided by telecommunications companies. One common example is cable modem or DSL, which can provide anywhere from .5 Mb/s to 5 Mb/s at $50 to $150 per month. Another example is a T1, which provides 1.5Mb/s for about $300 to $600 per month. Above this level, bandwidth generally becomes very expensive. In most locations, getting 10Mb/s of bandwidth can cost thousands per month.
Many talk about fiber (sometimes called FTTH/FTTC) but fiber to the building is not and will not be widely available for years. Fiber to the home or to the business promises to reduce the cost of bandwidth significantly. Nevertheless, it is very expensive to deploy and despite excited discussions for the last decade or more, progress remains slow. If you have it great, but do not assume it.

Same Buildings
By contrast, bandwidth inside of buildings (or campuses) is quite high because the costs of deploying it are quite low. Non technical users can easily set up a 1000Mb/s networks inside a building (aka Local Area Networks or LANs) for less than $1,000 installation cost with no monthly costs. Contrast this to the WAN, where the same bandwidth could cost tens of thousands of dollars per month.
The cost of deploying networks in buildings are low because there are minimal to no construction expenses. When you are building a network across a city, you need to get rights of ways, trench, install on telephone poles, etc. These are massive projects that can easily demand millions or billions of dollars in up front expenses. By contrast, inside a building, the cables can often by quickly and simply fished through ceilings (not the professional way to do it but the way many people do it in deployments).
A lot of discussion about wireless (WiMax, WiFi, 3G, etc) exists but wireless will not provide significantly greater bandwidth nor significantly better costs than DSL or cable modem. As such, wireless will not solve the expense and limitations of bandwidth between buildings. That being said, wireless absolutely has benefits for mobility purposes and connecting to remote locations that DSL or cable modem cannot cost effectively serve. The point here is simply that it will not solve the problem of bandwidth between buildings being much more expensive than bandwidth inside of buildings.

How Much Bandwidth Do IP Cameras Consume?
For the bandwidth consumption of an IP camera, use 1 Mb/s as a rough rule of thumb. Now, there are many factors that affect total bandwidth consumption. You can certainly stream an IP camera as low as .2 Mb/s (or 200 Kb/s) and others as high as 6 Mb/s. The more resolution and greater frame rate you want, the more bandwidth will be used. The more efficient the CODEC you use, the less bandwidth will be used.
For the bandwidth consumption of a Megapixel camera, use 5 Mb/s to 10 Mb/s as a rough rule of thumb. Again, there are a number of factors that affect total bandwidth consumption. A 1.3 megapixel camera at 1fps can consume as little as .8 Mb/s (or 800 Kb/s) yet a 5 megapixel camera can consume as much as 45 Mb/s.

What Does this Mean for my IP Video System?
Just like dealing with personal finance, we can now figure out what we can 'afford':
Between Buildings: We have .5 Mb/s to 5 Mb/s to 'spend'
Inside Buildings: We have 70 Mb/s to 700 Mb/s to 'spend'
IP cameras: Cost us 1 Mb/s each
Megapixel cameras: Cost us 5 Mb/s to 10 Mb/s each
Using these points, we can quickly see what combination of IP and megapixel cameras we can use between buildings or inside of buildings.
Inside of buildings, it is easy to stream numerous IP and megapixel cameras.
Between buildings, it is almost impossible to stream numerous IP and megapixel cameras.
Because of this situation, the standard configuration one sees in IP Video systems is:
A local recorder at each building/remote site. The local recorder receives the streams from the building and stores them.
The local recorder only forwards the streams (live or recorded) off-site when a user specifically wants to view video. Rather than overloading the WAN network with unrealistic bandwidth demands all day long, bandwidth is only consumed when a user wants to watch. Generally, remote viewing is sporadic and IP video coexists nicely with the expensive Wide Area Network.
The local recorder has built-in features to reduce the bandwidth needed to stream video to remote clients. Most systems have the ability to reduce the frame rate of the live video stream or to dynamically reduce the video quality to ensure that the video system does not overload the network and that remote viewers can actually see what is going on the other side. Generally, the live video stream is sufficient to identify the basic threat. In any event, bandwidth is generally so costly, especially the upstream bandwidth needed to send to a remote viewer, that this is the best financial decision.
Knowing how much bandwidth is available for DVRs and NVRs and how much bandwidth IP and megapixel cameras consume are key elements in planning and deploying viable IP video systems. Though this is simply a broad survey, my hope is that this helps identify fundamental elements in understanding the impact of bandwidth on IP video.

No comments: