A mobile computing environment typically involves groups of small, low-power devices interconnected through a mobile and dynamic network. Attempts to secure communication over these “ad-hoc” networks must be scalable to conserve the minimal resources of mobile devices as network sizes grow. In this project, the scalability of differing Group Diffie-Hellman security key generation implementations is examined. In theory, the implementation utilizing a data structure with the lowest theoretical run-time complexity for building the Diffie-Hellman group should prove the most scalable experimentally. A common modular framework was implemented to support generic Group Diffie-Hellman key agreement implementations abstracted from the underlying data structure and traversal mechanism. For comparison, linear, tree-based, and hypercubic Group Diffie-Hellman topologies were implemented and tested. Studies were conducted upon the results to compare the experimental scalability of each implementation to the other implementations as well as the theoretic predictions. The results indicate that the benefits of implementations with low theoretic-complexity are rarely experienced in smaller networks (less than 100 nodes,) and conversely implementations with high theoretic-complexities become unsuitable in larger networks (more than 100 nodes.) These experimental results match the theoretical predictions based on the mathematical properties of each implementation. Since mobile ad-hoc networks are typically small, less efficient, less complex implementations of Group Diffie-Hellman key agreement will suit most needs, however larger networks will require more efficient implementations.
Library of Congress Subject Headings
Public key infrastructure (Computer security)--Evaluation; Computer networks--Security measures; Data encryption (Computer science); Computer networks--Scalability
Computer Science (MS)
Department, Program, or Center
Computer Science (GCCIS)
Hagzan, Kieran S., "The performance of Group Diffie-Hellman paradigms: a software framework and analysis" (2007). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus