Research On Logic Circuit Failure Probability Sensitive Vector Search Algorithm

With the development of CMOS(Complementary Metal Oxide Semiconductor)manufacturing technology,the feature size of integrated circuits has been further reduced and chip performance has been steadily improved.At the same time,however,process disturbances,power supply noise,particle radiation and other problems causing circuit failures are becoming increasingly serious.The large-scale scaling of devices has increased the probability of random doping fluctuations or manufacturing inaccuracies during the manufacturing process,exacerbating process fluctuations in devices.Process fluctuations,ageing effects and external radiation are the main causes of failures in circuit components and pose a serious challenge to circuit reliability design.Effective fault-tolerance schemes to reduce failure probability are a hot topic of interest in industry and academia.Therefore,finding input vectors that are sensitive to the failure probability of a logic circuit can assist circuit designers in selectively reinforcing the circuit for fault tolerance,reducing the cost of fault tolerance and improving fault tolerance efficiency.The failure probability of a circuit varies considerably under different input excitations.If the sensitive vectors that cause the "failure probability shortfall" can be effectively avoided,and the sensitive gates corresponding to the sensitive vectors can be selectively fault tolerant and reinforced,the failure probability can be reduced at minimal cost.Existing methods have difficulty in balancing accuracy and efficiency in searching for sensitive excitations in vector spaces that are exponentially related to the number of circuit inputs.An efficient and accurate search for sensitive excitations is important for the reliability of integrated circuits and systems.In this paper,two solutions are proposed for the search of circuit sensitive vectors.1.Exploiting the efficiency of intelligent algorithm for finding the best,this paper proposes an improved Adaptive Cuckoo Search(IACS)algorithm for searching circuitsensitive vectors.A vector partitioning strategy is used to change the dimensionality of the initial input vectors,combined with a hill-climbing algorithm to improve the initial population quality,and an adaptive strategy to control parameters such as power law exponent,discovery probability and scaling factor to achieve fast and accurate search of circuit-sensitive vectors.At the same time,the circuit failure probability under specific vector excitation is calculated quickly and accurately using the correlation separation method,which effectively improves the efficiency of the algorithm.Experimental results show that the method proposed in this paper has higher accuracy and better efficiency than existing sensitive vector search algorithms.2.Using the idea of compressed vector space,this paper proposes a fast and efficient method for searching sensitive vectors in logic circuits.This method uses a fixed bit strategy to determine the worst input bit and a consistency strategy to improve the speed of the algorithm.Experimental results show that this method has a large advantage in terms of time consumption for searching sensitive vectors and has a high accuracy.For circuits with a small number of inputs,the method is more efficient in searching for sensitive vectors than the improved adaptive cuckoo algorithm.