Design And Implementation Of A Probabilistic Circuit Fault Injection Method Based On Monte Carlo

With the mature of deep sub-micron technology and the development and application of nanotechnology,VLSI(Very Large Scale Integrated,VLSI)circuit is getting more and more widely used in every aspect of life,to the development of national economy and the progress of science and technology is playing a more and more important role.The application of new technology and new process give the circuit chip integration greatly ascend,bring new opportunities for integrated circuit design at the same time,also pose a great challenge.Among them,the circuit reliability analysis and design based on nano-devices is one of the major challenges facing.At present,there are multiple ways of evaluating reliability of integrated circuits.Among them,the fault injection technique as a kind of evaluation method,got the extensive application of the researchers.Monte Carlo method as a kind of probabilistic fault injection method,it has the advantage of strong commonality,from small to large scale logic circuit can use it for testing and evaluation.And the method is often used to compare with other methods and as a reference standard for validating other methods.However,the method also has a more obvious drawback that the time to evaluate the reliability of the circuit is greater,especially when the circuit size is relatively large.Nevertheless,the Monte Carlo method is still a very valuable international reliability evaluation method.The traditional experience-based Monte Carlo method is more inconvenient in testing large-scale circuits.In view of this problem,this paper improves the access mode of the traditional Monte Carlo method to simplify the model by means of the convenience of the access and operation of the linked list,and the results of this paper are analyzed and compared with the results of other related literatures,which verifies the accuracy of the proposed method.In addition,this paper presents a Monte Carlo method based on adaptive strategy for some shortcomings of Monte Carlo method based on empirical value,such as the blindness of the number of samples and the difficulty of determining the error range of the test results.The adaptive method does not need to set the number of samples to be tested,but according to the test parameters and relative error,through self-learning to determine whether the test results converge,when its convergence,the result is measured logic circuit reliability.Experiments are carried out and compared with the empirical Monte Carlo method.It is proved that the Monte Carlo method based on the adaptive strategy is accurate and reliable.Moreover,compared with the empirical Monte Carlo method,this method can dynamically control the error range of the test results as needed.