Square-Root-Domain Realization of Single-Cell Architecture of Complex TDCNN
- PDF / 1,073,682 Bytes
- 20 Pages / 439.37 x 666.142 pts Page_size
- 39 Downloads / 150 Views
Square-Root-Domain Realization of Single-Cell Architecture of Complex TDCNN Farooq A. Khanday · Costas Psychalinos · Nisar A. Shah
Received: 28 March 2012 / Revised: 21 September 2012 / Published online: 11 October 2012 © Springer Science+Business Media New York 2012
Abstract Square-root-domain (SRD) CMOS analog realization of a single cell architecture of the complex Temporal Derivative Cellular Neural Networks (TDCNNs) is introduced in this paper. TDCNN initiates time derivative ‘diffusion’ between CNN cells for non-separable spatiotemporal filtering applications, where the input to the CNN is an image that changes over time. The evaluation of the performance of the complex SRD TDCNN cell has been done using the Cadence Orcad software with TSMC 0.18-µm CMOS process model parameters. The provided simulated results confirm the validity of the theory. Keywords Cellular neural network realizations · Continuous-time filtering arrays · Spatiotemporal filtering · Square-root-domain filtering · Analog signal processing
1 Introduction The revolutionary Analogic (a generic term for Analog and Logic) Cellular Computer paradigm is an attractive candidate for processing analog array signals. The core of this computer is a Cellular Neural Network (CNN), and an array of analog dynamic processors or cells (called neurons). At the same time, Analogic CNN computers F.A. Khanday () · N.A. Shah Department of Electronics and Instrumentation Technology, University of Kashmir, Srinagar 190 006, India e-mail: [email protected] N.A. Shah e-mail: [email protected] C. Psychalinos Physics Department, Electronics Laboratory, University of Patras, 26504, Rio Patras, Greece e-mail: [email protected]
960
Circuits Syst Signal Process (2013) 32:959–978
mimic the anatomy and physiology of many sensory and processing organs of humans with an additional capability of stored programmability. The concept of cellular neural network was originally proposed by von Neumann [28], and its electrical model was fully articulated by Chua and Yang in 1988 [3]; since then CNNs have been the object of a great deal of research work, concerning both theoretical studies and application-oriented circuit implementations. CNNs have been applied in various fields such as computer vision [4, 24], robotic control [19], and nonlinear modeling [16, 29], and are well suited for high-speed image processing tasks. The reported applications cover a much wider range of activities, such as motion detection, classification and recognition of objects, associative memory, solution of partial differential equations, statistical and nonlinear filtering [5]. The key features of neural networks are asynchronous parallel processing, continuous-time dynamics, and global interaction of network elements [14]. Despite many years of studies involving neural networks, it is only due to advancement in programmable hardware that actual implementation for study and its applications has become practical. There are many examples of neural network codes running on von Neumann
Data Loading...
