DocumentCode :
418109
Title :
Improved cell core for a mixed-mode polynomial CNN
Author :
Laiho, Mika ; Paasio, Ari ; Halonen, Kari
Author_Institution :
Electron. Circuit Design Lab., Helsinki Univ. of Technol., Finland
Volume :
3
fYear :
2004
fDate :
23-26 May 2004
Abstract :
This paper proposes improvements for a recently reported mixed-mode polynomial CNN cell structure. Like the previous cell core, the improved cell can be used to process input data that has been divided into blocks (layers), while keeping all input/output operations during processing local. The improved cell core uses current memories as means to convey data between different blocks. This way only one DA converter and polynomial circuit is needed in the cell and the current sources of the DAC can also be utilized in the ADC. In addition to gray-scale processing, the cell is capable of processing binary data so that the eight-bit gray-scale state memory can be used as eight one-bit local memories. The analog circuits of the design are composed of simple primitives which make the design well suited for implementation using modern silicon process.
Keywords :
analogue processing circuits; cellular neural nets; digital-analogue conversion; integrated memory circuits; mixed analogue-digital integrated circuits; ADC; DA converter; analog circuits; binary data processing; cell core; cell structure; current memories; current sources; gray-scale processing; mixed-mode CNN; modern silicon process; polynomial CNN; polynomial circuit; Algorithm design and analysis; Analog circuits; Cellular neural networks; Electroencephalography; Electronic circuits; Epilepsy; Gray-scale; Laboratories; Polynomials; Voltage;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on
Print_ISBN :
0-7803-8251-X
Type :
conf
DOI :
10.1109/ISCAS.2004.1328691
Filename :
1328691
Link To Document :
بازگشت