Re-Ordering of Users in the Group Key Generation Tree Protocol

Abstract Tree-based Group Diffie-Hellman (TGDH) is one of the efficient group key agreement protocols to generate the GK. TGDH assumes all members have an equal computing power. As one of the characteristics of distributed computing is heterogeneity, the member can be at a workstation, a laptop or even a mobile computer. Therefore, the group member sequence should be reordered in terms of the member’s computing power to improve performance. This research proposes a reordering of members in the group key generation tree to enhance the efficiency of the group key generation. Key Words : Protocol design, network security, authentication, group communication, agreement protocols, group key management. * 소속 직책: 백석대학교 정보통신학부 조교수논문접수: 2012년 6월 27일, 1차 수정을 거쳐, 심사완료: 2012년 7월 23일 1. Introduction Android phones and iPhone are used a lot by users. More than 70% of cell phone users are smart phones which are able to communicate with other smart phone users. Group communications are pervaded over the network such as video conferences and on-line chatting programs, games, and gambling. Security plays an important role in these instances of group communication. According to [11], user authentication processes and key distribution are just at the beginning of the group communication [11]. The CGK generation, on the other hand, takes a relatively long time to complete. For achieving a high level of security, the CGK should be changed after every user joins and leaves so that a former group member has no access to current communications and a new member has no access to previous communications [11] To improve the group communication efficiency, the CGK generation needs to be optimized by the improved key generation algorithm. Accordingly, group key agreement protocol focuses on the CGK generation. The function for generating CGK in the group key agreement is a modular exponentiation. In order to calculate the CGK using modular exponentiations, the adaptation of key trees is needed to reduce the computational overhead. Modular exponentiation is the computationally most expensive operation in TGDH [2]. The number of exponentiations for membership events depends on the number of