Design of very high speed CMOS fuzzy processors for applications in high energy physics experiments
D. Falchieri, A. Gabrielli, E. Gandolfi, M. Masetti · 2003
We faced the problem of VLSI fuzzy chip design because the processing rate of the fuzzy chips available on the market is too low for trigger applications in High Energy Physics Experiments. When we started, three years ago, we chose the 0.7 micron ES2 technology because it was the fastest one available at an accessible price. The paper describes chip architectures where each rule is processed in one clock period: 20 ns. Our goal was a processing rate of few hundred ns: it was possible to reach such a result only when the number of the inputs is less or equal to four: for two inputs the processing rate is 80 ns while for four inputs 320 ns because only the active rules are processed. For more inputs a genetic rule generator is used because it allows us to have a fuzzy system with a very low number of rules: for most applications the number of rules is about 10. The architecture of a 10 input fuzzy chip able to process a "genetic fuzzy system" is described. The above chips have been constructed and now we have redesigned the two input fuzzy chip using the 0.35 /spl mu/m technology of Alcatel Mietec to reach a higher processing rate.