Optimization Techniques For Executable Test Cases Generation And Prioritization For EFSM-based Systems

Software testing is very important in software development.More complex software brings new challenges to software testing.The model-based testing technique is more simple and efficient,which can guarantee the quality of test case generation and prioritization.The Extended Finite State Machine(EFSM)extends the variables and predicate conditions based on the finite state machine,which can help models to describe the data flow of software.However,this also brings the problem of state explosion and unexecutable result sequences.In addition,how to prioritize the execution of test cases based on the information contained in models to further improve the fault detection capability of test cases in regression testing is also a challenge currently faced.Based on this,this article focuses on how to efficiently generate feasible test cases for EFSM models and prioritize the execution of test cases.This paper proposes a new method of test case generation based on Monte Carlo Tree Search(MCTS)and Transition Executable Analysis(TEA).Meanwhile,this paper proposes a new method for ranking test cases by integrating the existing heuristic ranking algorithms based on random forest.To verify the effectiveness and feasibility of the new methods,this paper carried out in-depth experimental research on five classical EFSM models.The results show that the new method is significantly improved compared to the traditional methods.The main work of this article can be summarized into the following points:(1)A method for generating feasible test cases of EFSM based on Monte Carlo tree search is proposed.This new method contains a fusion algorithm base on TEA and MCTS.The search space can be greatly reduced while ensuring the executability of test cases.Experimental results show that the new method can effectively alleviate the state explosion problem in the process of generating feasible test cases.(2)Propose a method to sort test cases of EFSM based on random forest.The new method integrates 7 heuristic ranking algorithms through the random forest model in learn-to-rank to comprehensively utilize EFSM's multi-dimensional feature for sorting test cases.Experimental results show that compared with a single ranking algorithm,the new method can significantly improve the average fault detection rate of the test suite.(3)An experimental prototype system was designed.This system can implement the feasible test case generation and prioritization methods proposed in this paper.The system has a simple human-computer interaction interface and data management function.The system is more effective for the test case generation and prioritization of the EFSM model in actual use.