Research On Overhead Controllable Runtime Verification Framework Based On Predictive

As a new software quality verification method, software runtime verification is different from the traditional methods of software testing and model checking, by avoiding the problems of state explosion and incomplete test cases. And that makes the runtime verification technology more and more be noticed. However, due to runtime verification being based on the software system for real-time monitoring, monitoring behavior influence the efficiency of software in many aspects. Aiming at this problem, this paper proposed a new method to control the cost of software by predicting the behavior of software. Based on the processing and analysis of the data of the software history, the operation rules of the software are studied, and the future operational risk of the software is predicted. When the runtime verification on the target system operating efficiency of the effect of can't be ignored, the risk of the future operation of software can be used to guide the operation of the verification system for monitoring the behavior of their own for self-regulation, so as to maintain the normal operation of the software itself.The main work of this paper is as follow:1. On the basis of researching the exist runtime verification techniques, this paper proposed a predictive based overhead control software runtime verification framework. The framework improved the traditional runtime verification flow, and joined the means to improve the efficiency of the system. At first,the framework used PEDL and MEDL scripting language to instrument and collect software history data, carries on the analysis to the risk assessment of software in the future operation by using Markov Chains and Hidden Markov Model. Finally, control monitoring software behavior according to the result of risk assessment. And will these content be combined to the MaC frame.2. In this paper, the method of predicting the future risk of software according to the historical information of running software was proposed. Using Markov chain and hidden Markov model, solutions were given for the situations that all the states of the system that can be observed and the situations that some of the states are belong to the hidden part of the system, respectively. Firstly, the method of establishing the software behavior model according to the historical data was given. And Secondly, the method of calculating the risk of the behavior of the software in the future was given according to the model. And on the basis of this, this paper gave the strategy of monitoring and controlling the behavior according to the risk of software running. Mainly used the methods of code version control, and the real-time control of monitoring, to achieve the focus of monitoring, to ensure the the effect of software monitoring.3. On the basis of the above theoretical research results, this paper designed and implemented the prediction based cost control software runtime verification prototype tool PCRV. And the prototype tool were applied to a case. At last, a experiment were carried out to analysis the framework. It is proved that the runtime verification framework in this paper can improve the efficiency of the verification system distinctly.