Computing with Time: From Neural Networks to Sensor Networks

This article advocates a new computing paradigm, called computing with time, that is capable of efficiently performing a certain class of computation, namely, searching in parallel for the closest value to the given parameter. It shares some features with the idea of computing with action potentials proposed by Hopfield, which originated in the field of artificial neuron networks. The basic idea of computing with time is captured in a novel distributed algorithm based on broadcast communication called the Lecture Hall Algorithm, which can compute the minimum among n positive numbers, each residing on a separate processor, using only O(1) broadcasts. When applied to sensor networks, the Lecture Hall Algorithm leads to an interesting routing protocol having several desirable properties.

• Type: Journal
• Year: 2008
• Where: CJ
• Authors: Boleslaw K. Szymanski, Gilbert Chen
• Keywords: Artificial Neuron Networks, CJ 2008, Closest Value, Lecture Hall Algorithm