進化型アナログLSI

遺伝的アルゴリズムによる製造誤差への適応

抄録

We propose a concept of evolvable analog LSIs and apply it to Intermediate Frequency (IF) filters, which are widely used in cellular phones. The developed chip can correct discrepancies in the values of analog circuit components by genetic algorithms (GAs), which have attained (1) yield rate of 97% (2) 60% reduction of the filter area (3) 38% reduction of power dissipation, compared with AKM commercial products. When analog Integrated Circuits (ICs) and Large-Scale Integrated Circuits (LSIs) are manufactured, the values of the analog circuit components, such as resistors or capacitors, often vary from the precise design specifications. Analog LSIs with such defective components can not perform at required levels and thus have to be discarded. However, using our technique can improve yield rates of analog LSIs and merit from two more advantages, namely: smaller circuits and less power dissipation, which can lead to cost reductions and efficient implementation of analog LSIs. This approach could be applied to a wide variety of analogue LSIs.