The Internet and Post-Computing Technology

Abstract

We are rapidly approaching the time (if we haven't already reached it) when the majority of the things on the internet will not be computers, but things that contain computers. Such things will include cell phones and PDAs, but will also include refrigerators, automobiles, weather stations, agrucultural sensors, and light switches. When the number of things on the internet gets considerably larger than the number of people on the planet, it will become apparent that most of the traffic on the internet is between various things that aren't people. The first question we will address is why we would want such an internet--the examples usually given are unsatisfying, and we will dwell on less glamorous but more practical uses. We will then turn to the question of how we will build an internet of things. Our current techniques will not scale to the size we will soon be experiencing, nor will our current models allow the kind of reliability that will be required. Instead, we will suggest an approach that combines aspects of distributed computing and mobile code to allow such systems to self-organize and evolve without central planning, an approach that allows models that scale to sizes that are appropriate to the internet of the future