Group Secret Key Extraction And Capacity Analysis In Wireless Networks Based On Clustering Algorithms

Secret key generation using randomness of wireless channel has been a new alternative for ensuring security in mobile environments.Received signal strength(RSS)based point-to-point secret key extraction gains much attention due to its easier accessibility than other wireless channel feature.Compare to massive studies on point-to-point secret key extraction,the problem of using wireless channel feature to practically perform group secret key generation among multiple wireless devices remains a challenge.Therefor,this paper primarily focus on extracting the group secret key among devices(the number of devices is greater than two),and the main work can be summarized as follows:1)Firstly,three existing methods of group key generation are analyzed,including the method of differential group key extraction,the algorithm of group key generation based on maximum spanning tree,and the scheme of generating the group key by segment,and discuss their strengths and weaknesses respectively.Among them,the differential group key extraction method requires key negotiations repeatedly,and the group key capacity is not high.The group key extraction algorithm based on the maximum spanning tree needs to find a maximum spanning tree in a group which is difficult in the actual situation,and it is also time-critical to reduce weight of the corresponding edge when every one-bit key is generated.Although the algorithm of generating the group key by segment is more secure,but the delay of the key extraction process may eventually be longer.2)Secondly,a new single group secret key generation scheme is proposed.As to the disadvantage of differential group key extraction approach,a new group key extraction algorithm based on pairwise generation strategy is developed.According to the shortest length of local keys between the central node in the cluster and other nodes,the key is generated randomly using a uniform distribution,then the group key is extracted in a one-to-one manner in our framework.And it is proved theoretically that when the number of nodes in the group is greater than two,the performance of the proposed scheme is better than that in the literature and the complexity is lower.3)Since the decentralized distribution of nodes within a group in the realistic wireless network environment,there is a major flaw in the proposed pairwise generation group key extraction algorithm that the group key capacity will be severely limited due to large difference of local key capacity.To this end,a multi-subgroup hierarchical group key generation scheme is proposed,which can be decomposed into two problems: one is the node grouping,and the other is the protocol of group key reconciliation.The node grouping is discussed in chapter 5,and it consider how to divide the nodes in a group into multiple subgroups to maximize the signal to noise ratio between nodes in each sub-group,which can optimize subgroup key capacity.Whereafter,it is proved theoretically that the optimization model can be obtained by generating specified K clusters using efficient affinitypropagation(K-AP).A group key generation method based on clustering algorithm is further investigated in chapter 6,which can be adopted in the situation that the distribution of nodes within a group has an arbitrary topology.Then this paper prove that the proposed scheme can achieve the group key capacity improvement compared to the pairwise generation group key extraction algorithm,and verify the correctness of the theory by extensive simulations.