A Hybrid GRU-KBNN based Approach for Estimating Jitter in a Chain of CMOS Inverters

An efficient hybrid approach based on the gated recurrent unit (GRU) and knowledge-based neural networks is presented for prediction of power supply induced jitter (PSIJ) in the presence of multiple power supply noises in a chain of CMOS inverters. Also, an efficient method for generating an optimal training data set is presented. The proposed approach provides a reasonable accuracy and a significant increase in computational efficiency compared to conventional approaches.