Reliable computing with unreliable components
Project Leader: Jonathan Manton
Primary Contact: Jonathan Manton (email@example.com)
Keywords: fault-tolerant computing; novel architectures for computation; signal processing; signals and systems
Disciplines: Electrical & Electronic Engineering
Research Centre: Nonlinear Signal Processing Lab
Digital computation is currently based on logic gates, which in turn are made out of FETs. As the size of FETs shrinks, they become less reliable due to quantum effects and thermal noise. This motivates developing computational architectures built from unreliable components yet which compute reliably. There have been several approaches to this problem developed by others, including using error-correction coding. In addition to extending existing techniques, this project also studies novel architectures inspired by the brain.