Making Mistakes to Conserve Energy

Although human ingenuity has ushered in the digital age, we continue live in an analog world. Digital information consists of discrete values like 0 and 1, while analog is the continuum of those values and everything in between. Many familiar devices work by converting analog inputs from Nature, like radio waves or the images from a family vacation, into digital approximations which can be heard over a cell phone or viewed on a digital camera.

In the simplest terms, conversion from analog to digital is done by rounding the input to the nearest discrete number. The faster a digital device is, the faster it has to round the numbers, which requires more energy – unless the rounding doesn’t have to be perfect. Dr. Boris Murmann hopes to conserve energy needed for conversion by allowing devices to make more mistakes in the process. While less-precise measurements of analog input require less energy, they result in error-prone output that needs to be corrected.
Murmann is developing ways to detect and recover from the rounding errors with digital postprocessing. To do this, a “pilot” (a known second signal) is injected along with the main analog input. After converting everything to an imprecise digital version, the pilot is separated from the main signal and interpreted. The information retrieved from the pilot is subsequently used to correct any mistakes before generating the final digital output.

An OTL Research Incentive Grant helped fund the first step of this endeavor, formulating the algorithms necessary to correct conversion errors. However, work continues in Murmann’s lab toward the ultimate goal - developing a microchip capable of making analog/digital interfaces fifty times more energy efficient. With that much extra battery life, it’s going to be much harder to keep a teenager off the cell phone.

Boris Murmann
Assistant Professor of Electrical Engineering
“Mathematical Foundations of Next Generation
Analog/Digital Interfaces” (2004)