News
IP: past, present, and future

Eleven years later and we find ourselves in the same spot with analogue CCTV. There are still some people who believe analogue surveillance will remain dominant, but the recent decline in analogue market shares for 2009 and 2010 says otherwise. Still, even though we live in an all-digital consumer world, IP in security accounts for only a mere 25 per cent of the market. As someone who has been in the IP security space for 10 years now, this is one of the great mysteries to me.

With analogue’s days numbered, let’s look back at how we arrived here today and then take a stab at predicting what the future may hold.

It is hard to believe that 15 years have passed since the network camera was invented. That camera could do one frame per second in CIF resolution, and had an “incredible” one frame per 17 seconds at VGA. That is 600 times less powerful than cameras today, which can do multiple streams of 30 fps at HDTV resolutions. The earliest network cameras were obviously not implemented for security applications at that frame rate or resolution, but were perfect during their early days for Web attraction. Remember the first time you saw a Webcam? I’ll never forget that initial appeal.

So for the first five to seven years, IP video was mostly used for the Web or other non-traditional security uses. In 2001 I got my first taste real of IP camera functionality. I was at a startup IP environmental appliance manufacturer that leveraged network cameras to help IT departments secure their critical data centres. The IT security directors went outside traditional security companies and purchased these islands of physical security solely for their IT equipment. I was amazed at the quality then, and astonished that all you needed was a browser to pull up the camera feeds.

Then, in 2003, the world’s largest network equipment manufacturer took part in the largest IP security implementation at that time. They switched out 300 DVRs and used video encoders to convert 3,000 analogue cameras to IP streams. They also integrated these cameras with their enterprise software in several regional centres, which trickled up to one major central location. Network surveillance had arrived in a big way that year.

Fast-forward to 2011 and you will see several 100,000+ camera implementations worldwide. There are camera sensors in excess of 20 megapixels. There are cameras that can run various analytics at the edge (on-camera), like motion detection, people counting, tamper alarms and automatic tracking. The network cameras 15 years later have far surpassed analogue with respect to image quality, intelligence and scalability.

For instance, the APIs (Application Program Interface) are being standardized with a third party standards group called ONVIF. ONVIF-compliant cameras are being introduced by all the major manufacturers, indicating that it is becoming the de facto API standard. This will give end-users the advantage of buying products that are truly open and have a longer life in the field.

HDTV — a standard borrowed from the SMPTE and one that’s prevalent in the entertainment world — has become a major driver in security now that H.264 compression (yet another standard) has made higher resolution and frame rate in network video possible without bandwidth and storage baggage.

All these innovations together — image quality, intelligence, scalability, and standardization — will continue to drive down total cost of ownership of IP systems, priming the industry for rapid growth.

So what will happen during the next 15 years? One thing is for sure: innovation will continue, and at a fast pace. This is thanks mostly to “planned innovation,” known in IP circles as Moore’s Law — a doubling of capabilities every two years for the same cost. Network video innovation has actually surpassed Moore’s Law since 1996 and, if the Law holds true, it means that network cameras in 2026 will see almost another 600 times the capabilities that they have today.

Does this mean we will have terapixel sensors or cameras capable of running 10,000 frames per second? Maybe, but it could be that extra processing power is used to refine the camera’s output in a way that analogue ancestors could only dream of.

Additionally, since an IP camera by definition is a computer with an image sensor at the edge of the surveillance network, it’s logical that planned innovation with the processor will allow the camera to handle even more complex analytics at the edge. The camera will eventually become more and more independent of a central software server. This type of planned innovation opens up doors to brand new innovation from software developers and security directors.

The future looks bright for network cameras and I look forward to experiencing it firsthand.