Group Key Management In Ad Hoc Network

As a self-organized wireless network, Ad hoc not only contains wireless security threats, but also due to self characteristics: 1.fully distributed topology; 2.dynamiacal-changing group membership; 3.limited energy source, its security suffers more than any other wireless networks. Therefore, it has become one of the hottest research areas in recent years.In every layer of OSI network model, Ad hoc contains different security threats. Among them, the routing security and MAC access control are the most two popular issues focused on by nowadays researchers. For correct dilivery, digital signature, hash chain and the redundancy of routing scheme could be the answer. And to deal with node's bad and selfish behavior, grading system is always implemented to monitor them.However, all of the above measures are assumed that every node has their keys already. If no keys exist, none of them could be effective. In other words it also shows importance of key management in Ad Hoc network. In our scheme, we have analyzed the characteristics of Ad Hoc network. Moreover we have presented recent research works carried by scientists on group key management in Ad Hoc. We also try to find out the strengths and weaknesses of each scheme to show their application environment. And our conclusion is that none of these security plans could be totally suited to Ad hoc's dynamic topology. Our works:1. Analyze recent group key protocols and point out the weakness of each scheme. The schemes we have analyzed are LKH,LKH+,OFT,Cliques,KPS,RKP. For every protocol, we explain them from two aspects: 1. member joining; 2. member leaving. Meanwhile we introduced an improved plan for LKH+'s group merge. In LKH+'s group mergence, two sub-trees are just connected directly. The obtained new tree after merging may be an unbalanced one. This is not cost effective.2. Also we introduce a protocol which fits to Ad hoc's dynamic network topology. Three aspects we have achieved: 1. The group shared secret need only be changed when the leader of group leaves. When there is a not-group-leader leaver, every other node only needs to compute new keys——short period key locally. 2. Group leader is not selected randomly. Every node predicts their staying time and power. Through adding some weight to them, we obtain a grading value which indicates who will become the next group leader. By doing this we have decreased the frequency of changing shamir shared secret. 3. Access authentication has been classified into two aspects to speed join process caused by adverse environment.3. Finally, we proved our protocol can reduce the rekeying frequency by OMNeT++ simulation.