Study on leakage power estimation and reduction methodology of CMOS circuit

Zhi-qiang CHEN,Xiao-bo WU,Xiao-lang YAN
DOI: https://doi.org/10.3785/j.issn.1008-973X.2006.05.009
2006-01-01
Abstract:To reduce power leakage of deep sub-micron metal-oxide-semiconductor (CMOS) circuit in standby mode and search the minimum leakage vector (MLV) producing minimum leakage power, a linear programming (LP) model based on leakage power library was proposed. A modified genetic algorithm for MLV searching, which is based on the proposed LP model, was presented also. According to the leakage power value corresponding to individual state of each basic cell in the library, the total leakage power of the circuit was estimated on gate level. Using the LP model as evaluation function, this algorithm searched for MLV from the input vectors set by operations such as natural selection, crossover and mutation. Simulation results showed that the presented methodology could reduce the leakage power effectively. The modified genetic algorithm is convenient for implementation, and suitable for estimating and reducing the leakage power of Very Large Scale Integrated (VLSI) circuits.
What problem does this paper attempt to address?