Arithmetic arrays using cellular neural networks

This paper discusses techniques for using locally connected analog cellular neural networks (CNNs) to implement arithmetic arrays. These arrays are targeted at low speed low-noise applications where continuous power/speed trade-offs and lower slew rate during transitions are potential advantages. Specifically, we demonstrate that a CNN array using a simple nonlinear feedback template, with hysteresis, for each node, can perform arbitrary length binary addition with good performance in terms of stability and robustness. The processing speed can be controlled by changing the self-feedback value of the templates. We also propose a method for using the adder in binary multiplication.