| In this thesis, we study the performance behavior of mobile agents in the distributed search. The search problem is analyzed from two different standpoints: the single agent search and the multiple agents search. In the case of multiple agents search, we suggest three different algorithms and study their difference. We provide an architecture to implement a distributed search, which involves single and multiple agents as well as other components that guarantee agent coordination. Several experiments were carried out to measure the performance of the mobile agents to solve the distributed search problems under a variable network size, network topologies and number of agents. The test results of two different mobile agent platforms were also compared. The results indicate that the two mobile agent platforms have similar behavior but their performance varies with underlying implementation, and the multiple agents approach performs better than the single agent approach in large networks. Moreover, the hypercube topology provides the best performance among all topologies considered. |
