Ascenten Newsletter

Having trouble viewing this email? View in browser

Electronic Design
Embedded Software
Product Engineering

December 2012 / Issue #8

From the MD's desk

	I once owed a friend a 1000 dollars and it was my time to start returning the money. I didn't have the money so I convinced him that I would pay him a dollar every day for the next 1000 days and then stop paying no matter what. And he agreed. I started paying him but soon realized that my payment would fall short or be in excess of a few cents everyday. After 10 days I had managed to repay only 8 dollars. "Ha!", he exclaimed, "You're short by 20%.
I'm worried that with such a large error, after 1000 days, you would end up paying me too low – like about 800 dollars!" I asked him to bear patience and trust my knowledge of statistics. After 100 days I had managed 91 dollars. "Well", I said, "Now I'm short by only 9%." Days passed and my random payments continued. After 1000 days I had paid him a total of 972 dollars. "So here we are", I concluded, "I'm short by only 2.8%. And the pizza is on me. That makes us even." And my friend walked away happily. The above principle also works in the measurement world. When meters (for example energy meters or mass flow meters) measure and aggregate, individual measurements may be way off (noisy) – sometimes with an error as high as +/-200%, however as the aggregation happens the percentage error in the total diminishes to within an acceptable value. And the government regulators walk away happily. The same principle applies to cameras: Longer the exposure time, higher the aggregation of light at every pixel, and lower the percentage error on that pixel and hence as an overall effect, clearer the image. For instance, the Hubble telescope's exposure times ranges from minutes to days depending on the noise present and clarity required. I choose a simple way to explain how this effect works. Let's take averaging. When we average individual noisy measurements, the averaged result is comparatively much closer to reality (that is lower in error), than the individual measurements. Most meters, for example temperature, pressure etc. average individual measurements to display a more accurate result. In other words, averaging is a noise filter – it diminishes noise. This is well known in engineering. What is less known is that aggregating is similar to averaging, except unlike as in averaging, you don't divide after you aggregate and in this sense aggregating also behaves as a noise filter. But averaging or aggregating takes time. Which means that meters always trade between accuracy and time depending on which is more important. If both are important, the meter electronics needs low noise design techniques. Philosophically speaking, the longer you look, the better the observations. However, if you don't have enough time to look but you still need better observations, use better glasses. There are beautiful pieces of mathematics that prove what I've just presented in layman terms. The concepts involved are those of random variables, sum of random variables, independence and correlation, standard deviation, power spectral density and of course the holy Central Limit Theorem of statistics. Send your valuable comments and questions to meet@Ascenten.net.
Meet Kumar

Power Line Communication

Riyaz Panarwala (Embedded Software Engineer, Ascenten)

Power Line Communication (PLC) is a rapidly developing technology that utilizes electricity power lines for the high-speed transmission of data and voice services. Without using any new wires, PLC can bring very high speed connectivity to home, schools, hotels and many more places. Internet connectivity is inevitable today and offering a high speed outreach has been a challenge. PLC can solve this problem by providing at least 1 Mbps speed - 20 times faster than a standard phone/modem connection. There are various other applications that PLC can provide which includes whole-house audio, security & monitoring, smart homes, internet telephony, video communications, broadband over power line and many more.