A Cellular Neural Network (CNN) is a neural network model in which cells are linked only to neighboring cells. In image processing, a CNN can be used for noise reduction and edge detection. Small-World Cellular Neural Networks (SWCNN) are CNNs extended by adding a small-world link, which is a global short-cut. Although SWCNNs have better performance than CNNs, one of the weaknesses of the SWCNN is fault tolerance. Previously, we proposed multiple SWCNN layers to improve the fault tolerance of the SWCNN. However, as this only addresses termination failures it is not sufficient. In this paper, we propose a Time Stamp Voting method to improve tolerance of intermittent faults. This method is superior to Triple Modular Redundancy (TMR).