Watch out for ‘viewer creep’, in IP video systems, writes Frank Crouwel, Managing Director, NW Systems Group.
A well-known form of creep many of you will have heard of is ‘mission creep’ linked to humanitarian interventions in war-torn countries which escalate into, initially unintended, long-lasting military campaigns.
In the world of software coding, programmers talk about ‘Feature Creep’ as system features or capabilities are extended mid-way through projects. Meanwhile, project managers frequently bemoan ‘Scope Creep’, the process by which a project grows beyond its originally anticipated size or complexity.
This phenomenon generally occurs when new features are added to product designs that have already been approved, without providing equivalent increases in budget, time and/or resources. All examples of Creep have one thing in common; they are bad news for the organisations which are subject to them. They can lead to project failure, waste vast amounts of resources and in some cases they can be catastrophic, over-stretching delivery teams and even forcing businesses into administration.
In the IP video world there are two more specific areas of Creep to consider: Network Creep and more specifically Viewer Creep. Network Creep is a well-understood problem for IT and network managers as they add new services the corporate network. Think for example about the rise of IP telephony, the ever increasing amount of data we all transfer and the addition of IP Video Surveillance systems that increasingly run via corporate LANs. How do you ensure the IP Video system you are planning will not over-burden your network if it expands beyond its initial scope? A starting point will be the various design tools available from IP camera manufacturers, such as the Axis system design tool: http://www.axis.com/products/video/design_tool/
Key network-related considerations when designing IP video systems are:
1.Number of cameras you need
2.Number of streams from each camera
3.Image resolution (i.e. CIF, 4CIF, HD 1080p, 5 Megapixel etc.)
4.Compression type or mix of compression types (i.e. MJPEG, H.264 etc)
5.Recording continuously, only on events or during specific time periods
6.Number of hours per day that you need to record
7.Frame rates required
8.Complexity of image including amount of movement and light in the scene
9.Whether there is resource-intensive transcoding of video footage in the mix
10.And of critical importance, the number of Viewers that will use the system.
The Axis design tool enables you to play with camera model and configuration selections, ensuring that these setups adequately meet the requirements that you have identified in the initial design drafts.
As a rule of thumb if you are installing fewer than ten cameras then these will normally be able to operate across an existing 100 megabit network switch. One camera delivering high quality images at high frame rates can use up to three megabits per second of available bandwidth using H.264 as the compression format (and significantly more when using MJPEG). If you are deploying 15 cameras then you will almost certainly need to upgrade to a switch with a gigabit backbone. If you are recommending this upgrade, the server running the associated Video Management Software (VMS) must also have at least one gigabit network adapter.
Viewer creep
But the complexity does not stop there. We believe that there is a further key network consideration which is rarely factored in correctly. It is the amount of people viewing the video on a regular basis and what they are viewing the images on. If this is not properly assessed at the start of a project, it can have a negative impact on system performance very quickly. This is such a big issue that inside NW Systems we dub it Viewer Creep. Viewer Creep is becoming a bigger problem as video systems find wider appeal beyond security departments.
Up until a few years ago all CCTV cameras were viewed only by one department, normally within the confines of a monitoring centre or control room. But today IP video systems are delivering value to multiple departments. For example in a large warehouse where we installed a system recently, the video is being viewed by the head of IT, the warehouse manager, the food safety compliance manager and the security manager. Having access to images from the 13 cameras we have fitted there is of value to all of them. More than this, at least two now view images from their smart phones.
Viewer Creep is a real concern because it is often invisible to the integrator. Customers can add viewers by simply downloading the front-end desktop viewing software or mobile viewing software to a new device. It takes minutes and the integrator may not even know it has happened until they get a complaint that the network is slowing down or frames are being dropped as they view the live streams.
Each viewer in this instance creates up to 13 streams of video at the pre-configured frame rate and resolution so that could equate to a 40 Mbit/s additional bandwidth requirement for one extra viewer at the desktop. The advantage of viewing at the desktop is that no additional transcoding is needed to render the images onto the standard desktop so the 4CIF resolution image using H.264 compression can be displayed as it is streamed.
However, as is increasingly the case, if the new user wants to view images on his new iPhone then the image must be compressed further for viewing on this device and this creates a significant additional processing load on the server. This load is so significant that just one additional mobile viewer could force an upgrade of both RAM and CPU of the server. It could mean an upgrade to Dual Core 3.0 GHz CPU and 2GB of RAM costing hundreds of pounds for example.
The worrying thing is that none of the video management server calculators out there today give you the estimates of additional network bandwidth and processing power required to extend viewing of a site’s cameras to another person, so many installers and integrators miss this fact and fail to anticipate for it in their specification.
Experience tells us that Viewer Creep is on the rise as new IP video systems begin to deliver value to the organisations beyond standard security. Many more stakeholders are finding value in seeing images which show elements of their business in action. And naturally they want to view those images in high resolution on the move, on their mobile device of choice.
We work hard at the beginning of an engagement to understand who the stakeholders are in any new system and calculate the likely number of viewers and views based on department heads that are showing an active interest in the IP video project.
We always recommend configuring servers and networks not just for what is needed on day one, but what is likely to be required in the years to come. We also like to build in headroom of between 20-30% to allow for unplanned usage expansion. We have discussions with IT department heads to make sure they understand the implications of Viewer Creep and extending viewing capability to mobile devices.
By planning early you avoid network and processor performance issues which in turn impact on the quality and the efficacy of footage being viewed and recorded. And when you want to add more cameras or users to your system … always re-do your calculations.
