Digital Naval Gazing
Consider that Stream Processors SP8-G80 low power IP camera processor was delivered last month. It’s designed specifically for IP cameras, costs about $US59 wholesale and is likely to provide the basis for many next-gen IP cameras. The key issue is not huge price reductions. It’s likely to be quite the opposite. Quality IP cameras offering viable performance are going to take some research effort – effort customers should expect to pay for. The savings will come in reduced installation costs and greater operational flexibility. At the same time successful manufacturers will be able charge meaningful prices for advanced technology. The changeover to full IP cameras is also likely to mean that the pendulum of technological superiority will swing back to innovative R&D centres like the US, UK, Europe and Japan – markets currently under pressure from a deluge of me-too products offered by highly competitive manufacturing centres in Asia. Putting this to one side, however, what the switch to IP cameras is certain to mean is the presence of big, new competitors. It’s been speculated by industry analysists like Forward Concepts’ Will Strauss, that big DSP suppliers such as Texas Instruments, Analog Devices and Freescale Semiconductor will take advantage of new chips like the SP8-G80 but others like video and imaging DSP maker Cradle Technologies, and parallel processor outfits Ambric and Aspex Semiconductors will also muscle in on the act. According to Strauss, the new chip means that a number of companies are likely to hit the security market with massively parallel systems-on-a-chip specifically intended for video surveillance applications. As industry people well know, electronic security in all its forms is a stick in the mud. In Australia we all loved Advanced Security System’s brilliant new Geutebruck Multiscope eleven years ago (Advanced was the then distributor). But hard drive storage? Too small! Digital will never make the grade – it’ll be DAT all way… Those stick in the muds who think the electronic security industry will cling to CCD camera technology endlessly need to start thinking now about what they’re going to be selling in 5 years time – especiailly given the iSuppli projection that IP cameras will outstrip CCD gear in 2011. Consider that the world’s video surveillance market was recently projected to grow from $US4.9 billion this year (30 million cameras) to more than $US9 billion by 2011 (66 million cameras). That’s a compound annual growth rate of 13.2 per cent on dollars and just over 17 per cent in units sold. Major electronics manufacturers won’t be able to ignore a relatively immature (read margin-rich)market worth $US10 billion a year. Another element in the equation is the fact that all these security cameras are not going to be monitored by living breathing people. That’s just too expensive. It’s ironic that the recent big push in video surveillance has arrived hand-in-hand with reductions in the staff to monitor them. You needn’t be a rocket scientist to see where this is going to lead. Video analytics is coming, ready or not.Taking all this development into account there are a few ancilliary areas that are going to benefit. Storage is one area every self-respecting video surveillance integrator should be snuggling into. Now there’s a mature market with a static return on sales you can really rely on. Something else to think about is 10G Ethernet. Yeah sure – you don’t need it yet. But when developers start spilling out 10G over Cat-5 Ethernet hardware you might suddenly find that you do. Given most LAN/WAN networks remain incapable of offering end-to-end performance of 1G it might seem a bit rich to tout 10G. But if megapixel IP cameras wind up delivering the sort of performance they promise, the next Ethernet standard is going to need to offer something seriously special in order to keep up. Don’t believe me, though. In the last couple of months Intel has suddenly started talking about how it can run 10Gbps over standard Category 5 twisted-pair cables, for 100 metres with no repeater. I wonder why?