Simplicity in Electrical Engineering

January 27, 2012

The most elegant solution to a complex problem is often the most simple.

The concept is a fundamental tenant of mechanical engineering. Though many engineers only add complexity, the select few - the ones worth remembering - come up with an ingeniously simple solution.

In the age of micro electronics, it can feel like we’ve lost touch with this wonderful lineage. Physical mechanisms made of wound springs and metal gears have been replaced with silicon boards.

But electrical engineering has its own elegant simplicity.

A single resistor and capacitor can create a high-pass filter. Just four diodes can convert alternating current to direct current. And a single transistor opens the door to a mind-blowing list of possibilities.

Watching a presentation by Kwabena Boahen, I was impressed at what they achieved by applying the concept of simplicity to a traditionally complex field: the simulation of a field of neurons. Typically multi-purpose CPU and GPU chips are tethered together to run computations.

Instead Kwabena Boahen and his team created a simple neuron model in analog transistor and capacitor components. The result is fascinating, achieving huge density of neurons on a single chip, scalable to millions of neurons. The solution achieves moderately faster computation, massively cheaper cost, and many orders of magnitude lower power consumption than a traditional general purpose CPU and GPU with comparable output. All born out of specialization and simplicity.

I’m not advocating against complexity. Specialized solutions are inherently rigid, and come with their own set of quirks that may not make them preferable to more complex solutions. But when they do work, applied to a complex problem, a simple solution can be remarkable. And if you ask me, quite beautiful.