Date of Award


Publication Type

Master Thesis

Degree Name



Electrical and Computer Engineering

First Advisor

Jullien, G. A.


Engineering, Electronics and Electrical.




Cellular Neural Networks (CNNs) are massively parallel nonlinear locally connected analog cells; they are often targeted for use in image processing applications. An analog signal processor which (in majority of cases) does not require A/D conversion, occupies only a fraction of the area occupied by its digital counterpart. In image processing, it is possible to integrate an analog processor with each signal source (pixel) without leading to impractically low signal-source (or pixel) density. This makes possible the parallel loading or injecting of the input signal into the analog processor. Thus, sequential sampling at the output voltage (or current) of each signal-source is eliminated and a high degree of parallelism in signal processing is easily achievable. With the use of modified and creative algorithms, Cellular Neural Networks may also be used for digital arithmetic operations. For the same speed, these analog processors have lower slew rates compared to their digital counterparts which, in turn, leads to a lower generated noise. Using a Cellular Neural Network architecture, this thesis focuses on silicon implementation and experimental characterization of the building blocks for image processing and binary arithmetic applications using the MOSIS 0.5mum technology. These library cells are verified functionally at the layout level by conducting DRC, LVS, and post layout simulations. Sensitivity analysis is also carried out on a basic CNN cell in order to determine its tolerance with respect to expected variations in process parameters.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1997 .M38. Source: Masters Abstracts International, Volume: 39-02, page: 0565. Adviser: G. A. Jullien. Thesis (M.A.Sc.)--University of Windsor (Canada), 1998.