Deep Submicron Interconnect Timing Model with Quadratic Random Variable Analysis

Shrinking feature sizes and process variations are of increasing concern in modern technology. It is urgent that we develop statistical interconnect timing models which are harmonious with the current trend in statistical timing analysis flow. Although statistical model order reduction techniques have been explored, the statistical interconnect timing model has not yet been fully analyzed. In this work, we develop a novel algorithm and its corresponding analysis for the statistical interconnect timing model, using second-order statistical variations to model the non-Gaussian distribution effects. As this model is fully congruous with current statistical static timing analysis with the canonical model and does not require any Monte Carlo simulation analysis, performance is greatly improved. Experimental results show that the proposed closed-form quadratic interconnect timing model is within 0.0046% error of the corresponding Monte Carlo simulation.