Research On Test Data Generation Algorithms Of Combinatorial Coverage With Constraints

Electronic and information systems occupy an important position in many fields.The requirements of quality and reliability are very critical accordingly.Therefore,full-featured and full-state test has become an industry consensus.Full factor test is a more comprehensive and sufficient test method.However,sizes of full factor test data sets are usually very large,resulting in an explosion of test space in practice,which makes it very difficult to implement in actual tests.Combinatorial testing is a major black-box test method.Compared to full factor test,it only needs to detect interactions of any t parameters,which can not only guarantee small sizes of test data sets,but also guarantee the test comprehensiveness.The input of some electronic and information systems is timing and sequential.That is,the input must meet certain time and order requirements.There are various constraints among parameters and parameter values,such as combination constraints,order constraints and time constraints.It has become a hot issue in the field of functional testing to generate the smallest or smaller combinatorial test data sets that satisfy constraints while ensuring the test comprehensiveness.Research on the optimal or approximately optimal combinatorial coverage test data generation algorithms with constraints,which can not only meet the timing and sequential requirements and constraints of the input of electronic and information systems,but also ensure the test comprehensiveness and improve the test efficiency,is of great significance to improve the test coverage and product reliability of electronic and information systems.In view of this,this paper focuses on the test data generation of combinatorial coverage with different constraints.The main tasks accomplished are as follows.Aiming at the problem that the sizes of constrained coverage arrays are too large,the constrained coverage array generation algorithms based on particle swarm optimization are proposed.The optimization ability is improved by using coverage targets for initialization and global perturbation under the guidance of optimization rules.Two strategies,avoiding the selection of conflicting test cases and replacing conflicting test cases,are used to deal with the constraints respectively,and the latter strategy combines the local optimization and global optimization effectively.A size analysis method of 2-way constrained covering arrays based on tabu edge decomposition is proposed,which can verify practical feasibility of sizes of generated constrained covering arrays and simplify thegeneration process when used to generate constrained covering arrays.Experimental results show that,the constrained covering array generation algorithms based on particle swarm optimization can obtain the optimal constrained covering arrays under different coverage strength for the moment.Aiming at the problem that the existing description methods cannot describe order constraints involving specific input steps,a description method of order constraints based on clocked computing tree logic(CCTL)is proposed.Constraints involving specific input steps can be described by using the scope of temporal operators and the constraint description capability is enhanced.The verification method of constraint consistency of test data is given.Aiming at the problem that the test data of respective coverage of value combinations and value sequences has low testing efficiency when multiple parameters are input continuously and simultaneously,lacking of coverage criteria and generation algorithms of simultaneous coverage of value combinations and value sequences.Extended covering arrays are proposed.By combining two coverage criteria of value combination coverage and value sequence coverage,the completeness of coverage criterion is improved.Then,extended covering array generation algorithms based on the particle swarm optimization algorithm and the random algorithm are presented.The experimental results show that the particle swarm optimization algorithm can generate extended covering arrays with smaller sizes.Extended covering arrays of combinatorial strength and sequential strength of two reduce the sizes and improve the test efficiency compared with the test data of respective coverage.Aiming at the problem of incomprehensive testing caused by the low coverage of neighbor input time point combinations of random test data,when a single parameter is input continuously,neighbor covering arrays are introduced for the testing.When multiple parameters are input concurrently,concurrent input neighbor covering arrays are proposed,and the test data sets are generated based on the particle swarm optimization algorithm and the random algorithm.The experimental results show that,neighbor covering arrays can cover all the neighbor input time point combinations;concurrent input neighbor covering arrays can cover the time point combinations of two neighbor inputs of the same parameter and the time point combinations of any two parameters at the same input.The coverage of neighbor input time point combinations is improved and the test comprehensiveness is improved.An automatic test environment is built.Systems under test are abstracted into virtual I/O models which are able to accurately describe the functions.Virtual models are usedto provide expected test results.During the testing,test data are energized to systems under test from external interfaces while simultaneously energized to virtual models.Test results of systems under test are collected from external interfaces and compared with expected test results from the output of virtual models.Functional errors are defined as the difference between functional requirements and implementations,and the results of comparison are used to judge whether the systems under test have functional errors.Test data are used to practical tests in the automatic test environment,and the effectiveness and feasibility of the algorithms are verified.To sum up,this paper investigates the generation of more efficient combinatorial coverage test data according to different timing and sequential requirements and constrained requirements of input.It has achieved the goal of ensuring the test comprehensive and improving test efficiency.