Fault Tolerance in Analog VLSI: Case Study of a Focal Plane Processor
Biological systems provide good architectural models for information processing hardware. Difficult problems in machine perception and complex motor control are solved in a natural way by energy efficient and robust neural systems. Hopfield in his seminal paper  on physical systems with emergent computational abilities envisioned a new breed of integrated circuits that could implement such systems and would be much less sensitive to element failure than present day computers. Analog VLSI is a technology suitable for the implementation of synthetic neural systems [2, 3] on silicon.
KeywordsTracking Error Fault Tolerance Bias Current Voltage Difference Illumination Intensity
Unable to display preview. Download preview PDF.
- A.G. Andreou and K.A. Boahen, “Synthetic neural systems using current-domain signal representations,” Neural Computation, vol.1, 4, 1989.Google Scholar
- A.G. Andreou, K. Strohbehn and R.E. Jenkins, “A proposed scheme for the analog VLSI implementation of the Hassenstein-R.eichardt motion detector”, Presented at the 1989 Neural Networks for Computing conference. Snowbird; also Technical Report JHU/ECE-88/07, The Johns Hopkins University.Google Scholar
- S.P. DeWeerth and C.A. Mead, “A Two-Dimensional Visual Tracking Array”, Proceedings of the 5th MIT Conference on Advanced Research on VLSI, J. Allen and F.T. Leighton eds. pp. 259-275. 1988.Google Scholar
- S.A. Kontogiorgis and A.G. Andreou, “Mathematical Analysis of an Analog VLSI Visual Tracking Array”, Technical Report JHU-CS-88/17, Johns Hopkins Univ.; S.A. Kontogiorgis and A.G. Andreou, “Performance Analysis of an Analog VLSI Light Beam Tracking Array”, Technical Report JHU-ECE-89/04, Johns Hopkins Univ.Google Scholar
- S.A. Kontogiorgis and A.G. Andreou, “A Spatial Mean and Median Filter in Analog VLSI,” Proceedings of the 1989 Conference on Information Sciences and Systems, Baltimore, Maryland, 1989.Google Scholar