The Other Side of Power Conservation

I must admit that I was late to come around to the green revolution. I wasn't convinced that the human contribution to global warming was as great as it now seems to be until a couple of years ago. I thought that perhaps the power companies in advertising and promoting conservation were just trying to make more profit from scarcity - after all, it's easier to promote conservation than it is to build another power plant, especially when you know that demand is going to increase no matter what. Moreover, ROI for those in the power-generation business is much better in the former case. Cynical, me? Maybe.

And yet I am very conscious of the need to conserve power because I work in mobile. Everything we do depends upon batteries, and batteries often go dead at the worst possible time. Advances in battery technology come relatively slowly. So it makes sense to do whatever we can to conserve energy in the mobile world, on both the supply and demand side of the equation. As far as supply goes, we are still seeing some progress in battery technology itself, and we can use novel technologies like supercapacitors to even out demand, making the battery last a bit longer between charges. On the demand side, we can use more power-efficient semiconductor technologies, better chip designs that shut off unused circuits and are otherwise power-friendly, and the power-saving wireless protocols common to (if not essential in) all major systems.

But I've found it curious that this meticulous devotion to energy conservation in the mobile world doesn't often carry over to the other side of the link, which is powered from the AC mains (I just love the sound of the British terminology here). Case in point: a new power-over-Ethernet (PoE) standard, 802.3at, is almost done, destined to replace the current 803.af. 803.af is limited to 12.95 Watts at the end of a 100-meter link, which is enough to run VoIP handsets, security sensors, and even most current wireless LAN access points. But many products in the latest generation of 802.11n APs draw more than what .3af allows, and thus require either new PoE switches or new power injectors, some of which are today proprietary but all of which (in new products, anyway) will eventually be compliant with the upcoming .3at. Some network managers I've spoken with have .3at in their plans, and that makes sense - .3af is clearly being phased out, especially given the power challenge posed by .11n APs.

So I was surprised to learn recently that Siemens has a couple of dual-radio, 3x3 MIMO (300 Mbps raw throughput each) APs that they claim to be compatible with .3af. They offered me some lab time with one of the APs, and I ran some benchmark tests while powering both .11n radios in the AP from a four different .3af-compliant PoE switches and power injectors. It really worked, and really well - performance was excellent. You can read my full report here if you're interested (and you can find it in German here as well). This is quite an achievement (no, they wouldn't tell me how they did it), and I've heard from a couple of other players in enterprise WLAN systems who have advised me to stay tuned. Regardless, it can be done, and I'm encouraging all product developers to be creative when it comes to power - both supply and demand, and both mobile and fixed.

Apart from the obvious benefit of being a drop-in replacement for older .11a/g APs, thus minimizing upgrade costs, there's another real benefit here - it's time to think green even when we don't have to. Even when we have access to AC power, we should place minimizing power consumption at the top of our list of product requirements, whether we're design engineers or end-users. If saving the planet isn't enough for you, then think about this: sure, upgrading to new PoE equipment is expensive, but buying power we don't really need to run equipment that consumes more than it needs to is perhaps an even greater expense over the useful life of a given installation. And I think electricity is going to get much more expensive in the future, so the incentive to conserve is always there regardless.

Acknowledgement (the opposite of a disclaimer): As was noted in the Siemens press release, this work was funded by them. However, the tests were specified by me, conducted by me, and the Tech Note is 100% mine. It's hard regardless to fake test results. One side note, a couple of people have asked if I measured the actual power available on the cable in each case. No, I didn't, and it is possible that more than 12.95 Watts could be present. But Siemens' claim is that they'll work with .3af, and devices meeting that spec cannot count on more than 12.95 Watts. So, I thought that using four different, commercial, off-the-shelf (COTS) power sources would more or less prove the point. And if by chance these products all put out more than 12.95 Watts, and Siemens were to depend upon this, well, they'd have a whole lot of unhappy customers, lots of returned products, lawsuits, and the like. And I don't think they'd go to a lot of trouble to promote a product that would ultimately be just that - trouble.