Novel arithmetic implementations using cellular neural network arrays.

Date of Award


Publication Type

Doctoral Thesis

Degree Name



Electrical and Computer Engineering


Engineering, Electronics and Electrical.



Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.


The primary goal of this research is to explore the use of arrays of analog self-synchronized cells---the cellular neural network (CNN) paradigm---in the implementation of novel digital arithmetic architectures. In exploring this paradigm we also discover that the implementation of these CNN arrays produces very low system noise; that is, noise generated by the rapid switching of current through power supply die connections---so called di/dt noise. With the migration to sub 100 nanometer process technology, signal integrity is becoming a critical issue when integrating analog and digital components onto the same chip, and so the CNN architectural paradigm offers a potential solution to this problem. A typical example is the replacement of conventional digital circuitry adjacent to sensitive bio-sensors in a SoC Bio-Platform. The focus of this research is therefore to discover novel approaches to building low-noise digital arithmetic circuits using analog cellular neural networks, essentially implementing asynchronous digital logic but with the same circuit components as used in analog circuit design. We address our exploration by first improving upon previous research into CNN binary arithmetic arrays. The second phase of our research introduces a logical extension of the binary arithmetic method to implement binary signed-digit (BSD) arithmetic. To this end, a new class of CNNs that has three stable states is introduced, and is used to implement arithmetic circuits that use binary inputs and outputs but internally uses the BSD number representation. Finally, we develop CNN arrays for a 2-dimensional number representation (the Double-base Number System - DBNS). A novel adder architecture is described in detail, that performs the addition as well as reducing the representation for further processing; the design incorporates an innovative self-programmable array. Extensive simulations have shown that our new architectures can reduce system noise by almost 70dB and crosstalk by more than 23dB over standard digital implementations.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2005 .I27. Source: Dissertation Abstracts International, Volume: 66-11, Section: B, page: 6159. Thesis (Ph.D.)--University of Windsor (Canada), 2005.