A flexible simulation framework for distributed algorithms

Resource sharing and Routing are two fundamental issues in the distributed computing. Resource sharing and coordination arc achieved through distributed algorithms such as mutual exclusion, leader election, termination detection, and consensus and related problems, and Routing is achieved through routing algorithms. In the last several decades, a large number of distributed algorithms have been proposed. However, compared to the performance studies conducted for routing algorithms (using simulators such as NS2 and OPNET), very little work has been done in the performance study of distributed algorithms. More importantly, we could not find a comprehensive simulation framework similar to NS2 and OPNET that could facilitate the design and systematic performance analysis of distributed algorithms. This observation is the prime motivation for the contribution in this thesis. There are three main contributions of this thesis: (i) A comprehensive simulation framework to study distributed algorithms has been proposed, designed, and implemented. The simulator is expected to be available for public use; (ii) Two hypotheses have been formulated: first states that network topology has a fundamental influence on the performance of distributed algorithm and second says that the fairness and performance of distributed algorithms are two competing metrics; and (iii) A simulation study proves these two hypotheses.