A framework to study the performance of the group mutual exclusion algorithms

Group mutual exclusion problem generalizes the classical mutual exclusion problem, a fundamental problem in concurrent programming. It arises in applications involving sharing resources such as memory and data. In group mutual exclusion, a process requests for a “forum”; processes requesting the same forum may access the critical section simultaneously. Several algorithms have been proposed for the group mutual exclusion problem, but very few studies have been conducted to compare the performances of these algorithms by means of execution on actual machines. Besides the studies conducted have been a mere one-on-one comparison. Also, there exists no testing environment that accommodates multiple algorithms and compare their executions. This work aims at testing the performance of group mutual exclusion algorithms extensively by executing multiple such algorithms in a test framework. We propose to build an automated test framework to execute these algorithms, both individually and collectively under various experimental setups and observe their performances graphically using several performance metrics. Our experiments would constitute several collective comparison studies of algorithms along with replicating a few one-on-one comparison experiments from the literature. To use the algorithms into our framework, we intend to translate them from pseudo codes to source codes. The aim is to eventually creating a repository of these source codes such that they could be used for other applications besides our framework.