Architecture and Design of 1-D Enhanced Cellular NeuralNetwork Processors for Signal Detection

  • Authors:

  • Michelle Y. Wang;Bing J. Sheu;Theodore W. Berger;Wayne C. Young;Austin K. Cho

  • Affiliations:

  • Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089-0271;Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089-0271;Department of Biomedical Engineering, University of Southern California, Los Angeles, CA 90089-1451;Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089-0271;Department of Electrical Engineering, University of Southern California, Los Angeles, CA 90089-0271

  • Venue:
  • Analog Integrated Circuits and Signal Processing - Special issue: cellular neural networks and analog VLSI
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

One-dimensional cellular array processor architectureand design for neural-based partial response (PR) signal detectionare presented. Analog parallel computing approaches have manyattractive advantages in achieving low power, low cost, and fasterprocessing speed by its uniquely coupled parallel and distributedprocessing nature. In this paper, we describe the maximum likelihoodsequence estimation (MLSE) algorithm for PR signals, the enhancedCellular Neural Network (CNN) processor array architecture torealize the detection algorithm, and system performance evaluation.Analytical models and simulations on a design example of thedetector have been employed to demonstrate the advantages ofthis scalable VLSI architecture. A processing rate of 265 Mbpswas achieved for a prototype detector on a silicon area of 5.14mm by 5.81 mm is a 1.2 µm CMOS technology. Theprocessing rate can be beyond 1Gbps if it is implemented in thesame amount of silicon area by using 0.5 µm CMOStechnology. Such promising results clearly demonstrate the abilityto meet the needs in future high speed data communication byVLSI realization of maximum likelihood sequence detectors basedon the enhanced cellular neural network paradigm.