Wireless Sensor Networks -- ICT4D in embedded form ?

One technology that has fascinated me ever since I heard about it was Wireless Sensor Networks. It promises to reduce both cost as well as power consumed for computation. The technical challenges it offers (mainly in the field of routing protocols) are good enough to capture any self-respecting Computer scientist's attention. The amount of research going into this is a good indication that the technology holds tremendous commercial promise, with big players like Intel, Freescale, Texas Instruments vying to get a grip in the market.

The good thing about WSNs is that they are capable of automating various monitoring tasks which otherwise would have to be done by people. They are, being machines, are not susceptible to errors and falling asleep (unless the battery runs out!), and consequently, a highly reliable system. The standard that is presently the most popular is the IEEE 802.15.4, which takes care of low-level things and the Zigbee standard, which is a higher level thing based on the 802.15.4. A properly configured 802.15.4 device can run for months on one set of batteries, which makes it attractive where power is scarce and expensive.

Unlike regular computers, which have one CPU, the network behaves like a giant distributed computer, and hence designing algorithms is an interesting task. To cut up a big task into many small ones, distribute them over many small computers, then collating the results, and to do this reliably and efficiently, are big issues in the research community nowadays.

So, where does this fit into ICT for Development ? we don't really care about complex algorithms, but the attractive feature is low power consumption, and cheap cost of computation. There are competitive solutions which cost less than a thousand rupees per node. The applications are quite a few : Intelligent agriculture, Forest fire detection system, animal monitoring systems, to name a few. These are things that are done manually and with large inefficiencies in the process. Lack of regular, reliable information is the main issue, and that is what these networks address.

At AllGo Embedded, we have been working on some interesting applications which we hope will make a difference in the common man's daily life. More on this matter, as and when progress (or patents ;) happens.

LED lighting : when rather than if !

For the past few months, I have been looking at alternate lighting using LEDs. These devices show tremendous promise with regard to power saving, ease of use, and durability. So, thought would write a post for spread of info and concretizing my own readings.

LEDs, especially white LEDs are coming in big time into the market, and you can see them in all sizes and shapes in small flashlights and mobiles phones instead of flash bulbs. The biggest volume market right now is for low wattage(less power consuming) LEDs and Chinese goods are simply flooding the market. However, there is a silent background revolution happening, led by Philips, Cree, Nichia and Seoul Semiconductor. There is also an Indian company in the fray, based in Hyderabad. These and many other companies are trying to bring out high power LEDs which have direct applications in home and architectural lighting. These are rated in the range 1 Watt to 5 Watt. In contrast, the LEDs are found in your phones and flashlights are in the in the 200 milliWatt or less range, around one-fifth of the big guys.

These light sources are among the most efficient light sources available today, with efficiencies comparable with Compact Flourescent Lamps(CFL). However, most CFLs do not add the power consumed by a choke present on every lamp, and hence are usually less efficient than claimed. The unit of brightness is called lumen, and both LEDs and CFLs have an efficiency of around 50 lumen/Watt. In contrast, normal incandescent bulbs have an efficiency of 15 lumen/Watt! This implies you can replace a 100 Watt bulb with a 30 Watt CFL or LED and not perceive a difference in brightness. However, the light emitted by normal bulbs is more pleasing to the eye when compared to the other two. Therefore, you can see amber coloured CFLs and LEDs, neither of which have made a big dent in the market as of now, but promise to.

If CFLs are as good as LEDs, then why the whole fuss of typing out a whole post on them ? LEDs have many advantages, despite the fact that those in the market are only as efficient as CFLs.

• Prototype LEDs are available in labs of the above mentioned companies which are as efficient as 100 lumen/watt. However, most of these are sub 1 Watt category as of now. With all companies scrambling to outdo each other, this situation is likely to be rectified very soon.
• LEDs are solid state lighting devices, which means that they have no moving, breakable parts unlike either CFLs or bulbs. This means they can be used in more extreme places and applications.
• They are made from established manufacturing methods which make all our computer chips so cheap, which means at large volumes, the cost of lighting will be negligible. (Note: companies might keep cost high initially to recover cost. But once the Chinese get their hands on the technology, it should come down :) Evidence is available in the cost at which you get cheap LED based chinese goods. )
• These lights can be dimmed to suitable requirements, which is not an option on the CFLs. Bulbs can do this, but they do not even figure in the discussion.
• For spot lighting applications, where the light is required in only a particular area (street lights), LEDs are more suitable than the conventional tubelights (Sodium vapor lamps are not considered, since they are the most efficient lighting solutions with efficiencies of 150 lumen/Watt, but they are high voltage lights, and not used everywhere.) since they have a small angle beyond which the light output is almost zero. This means that all the light is focused onto a small area, unlike tubelights which radiate light every which way, which is essentially a waste of light.
• These run on DC current, which means that they can run off batteries in areas which do not have access to grid power. With suitable circuits, they can even run on AC current. These are cheap circuits and do not require as much circus work as running CFLs on DC, which require another kind of light itself.
• The (claimed) lifetime of LED lights is around 10,000 hours, which works out to be close to 10 years of operation. In comparison, CFLs have an average lifetime of 5-7 years, and forget about bulbs.
• The size of CFLs increases significantly with wattage increase, not so with LEDs.

My own tests with Kwality India's 1 Watt LEDs have been very promising. From a technical perspective, these babies require constant current rather than constant voltage, (which is what all our wall sockets and adaptors provide) but this issue has been solved as well, in a cheap way without resorting to expensive LED driver chips. There is a loss in efficiency, but not enough to give anyone sleepless nights. 4 LEDs and a driver circuit drawing a total power of around 5 watts gives enough light to illuminate a 40ft x 10ft area with reasonable amount of light, comparable to a 40 watt tubelight. These tests were subjective, but we can be sure that tests with a light meter won't be too far off as well. These lights will eventually find their way into streetlights in Timbaktu. Hopefully, very soon, will write about it when it happens.
So, why does not anyone as yet have LED lights in their house ? Packaging, lack of awareness, high prices (low volume :( ) all have contributed to this. Hopefully, this will change soon.