A Parallel Genetic Algorithm Based on Hadoop MapReduce for the Automatic Generation of JUnit Test Suites

Abstract

Abstract-Software testing represents one of the most explored fields of application of Search-Based techniques and a range of testing problems have been successfully addressed using Genetic Algorithms. Nevertheless, to date Search-Based Software Testing (SBST) has found limited application in industry. As in other fields of Search-Based Software Engineering, this is principally due to the fact that when applied to large problems, Search-Based approaches may require too much computational efforts. In this scenario, parallelization may be a suitable way to improve the performance especially due to the fact that many of these techniques are "naturally parallelizable". Nevertheless, very few attempts have been provided for SBST parallelization. In this paper, we present a Parallel Genetic Algorithm for the automatic generation of test suites. The solution is based on Hadoop MapReduce since it is well supported to work also in the cloud and on graphic cards, thus being an ideal candidate for high scalable parallelization of Genetic Algorithms. A preliminary analysis of the proposal was carried out aiming to evaluate the speed-up with respect to the sequential execution. The analysis was based on a real world open source library.

Efficient and Scalable Collection of Dynamic Metrics Using MapReduce

“…Sajnani et al [21] implemented a MapReduce based parallel algorithm for code clone detection that is capable of scaling to thousands of projects. Geronimo et al [22] proposed a parallel genetic algorithm that uses MapReduce to automatically generate JUnit test suites. Bianculli et al [23] exploited the MapReduce framework to check specifications expressed in a metric temporal logic with aggregating modalities (over large execution traces).…”

Parallel Bat Algorithm Using MapReduce Model

“…Formally, this problem can be framed as finding parameter 12 { , } X x x  where, {0,5} i x  that minimizes the following equation:…”

“…Keco and Subasi [14] proposed Model 2 and compared it with model 1 for same one max problem implementation. Then these models were used for various problems like Job Shop Scheduling Problem [3], automatic generation of JUnit Test Suites [12] etc. Lin, et al [8] scaled Modified Cuckoo Search using MapReduce Architecture and evaluated it on Griewank function, Rastrigrin function, Rosenbrock function and Sphere function.…”

Research on Parallel Particle Filtering Target Tracking Algorithm Based on Hadoop

“…Particle filter [14] is by looking for the random sample which spread in a group of state space to approximate probability density function ,which is a process to use sample mean instead of integral operation for obtaining minimum variance estimation and these samples is called "particle" [15].…”