A self stabilizing overlay networking algorithm using geographical segments for linearization in distributed computing

Abstract

This research proposed a new algorithm for building overlay distributed computing network by using geographical position of the nodes as consideration when building its topology. This research is done to prove the scalability of the proposed algorithm, to analyze the limitation of simulating the algorithm in terms of resources (memory) usage and to evaluate the degree of decentralization of the network which implementing the algorithm. A computer simulation is programmed to run and test the simulated network by analyzing how number of nodes affect the linearization process (in terms of number of cycles). The simulation also is used to analyze how many messages are passed during the linearization process to show limitations of simulation process. And lastly, the simulation also is used to calculate the degree of decentralization of the network. From the results, it can be concluded that the algorithm is scalable regardless the number of nodes. From another test, it was found that as number of nodes increases, number of messages passed among them collectively across the network increase exponentially which make it almost impossible to run the simulation of millions of nodes on a single machine. From the test also, the behavior of the network can be observed as number of supernodes increase, the network linearizes faster but approaching a limit.