Analog implementation of TDCNN single-cell architecture using sinh-domain companding technique
Nasir Ali Kant, Mohamad Rafiq Dar, Farooq Ahmad Khanday, Costas Psychalinos · 2016
Temporal Derivative Cellular Neural Network (TDCNN) is an important class of neural networks. These networks find a lot of application in real life mostly in the real-time image processing. However, the main challenge is to implement this network in hardware. Therefore, in this paper, sinh-domain realization of single cell architecture of TDCNN which forms the only building block of complex TDCNN is introduced. The design offers the advantages of; a) low-power operation, b) electronic tunability, c) grounded components, and, d) Class AB nature. The functioning of the cell has been verified by simulation results achieved through HSPICE simulation tool employing CMOS 0.35 m process.