Identification of most critical paths using sparse matrix in software testing

摘要

A software code may practically consist of many functions or subroutines with loops and branches. Also, the quantification of cyclomatic complexity is tedious, when the size of the software code is too large. Practically, size of the code for a real time application is generally large and dd-graph generation and maneuverability are tedious. Exhaustive testing is impractical and therefore it is not feasible to test all possible paths in the flow graph. Moreover, it is difficult to identify the uncovered portion, identification of most critical paths and hence unable to test all critical components in the software code. The present investigation focuses to display decisions along test paths in dd-graph of the software code using sparse matrix approach drawn through MATLAB biograph object module. In this approach, the concept of dd-graph (decision-to-decision graph) is taken from control flow graph of the software code by joining decision to decision. An identification of most critical paths and test case generation are done using artificial bee colony optimization. To avoid the testing complexity, the sparse matrix approach is used to demonstrate the most critical paths and display of the dd-graph using biograph object module by initializing the edge-node relationships is presented. The gcov code coverage analysis generates branch percentage probability as coverage summary which is considered as edge weights of the sparse matrix. This present approach is tested for the benchmark problem of "finding roots of the quadratic equation" software code.