A dynamic and multi-level key management method in wireless sensor networks (WSNs)

Abstract

Wireless sensor networks include a set of ultralight sensor nodes with limited energy, low storage capacity, and constrained processing power. Security is a challenging issue in these networks. One approach to maintain security in these networks is key management. When designing key management schemes, the main challenge is to achieve the acceptable security level and manage resources, especially energy. Today, many key management methods use authentication mechanisms. However, these methods are faced with different challenges due to the limited computational capability of nodes and high energy consumption in the network. In this paper, we present a dynamic and multi-level key management method for homogeneous wireless sensor networks. The proposed key management approach uses an authentication mechanism based on message authentication code (MAC), which is calculated by the shared symmetric keys between the transmitter and receiver nodes. In our scheme, the network is divided into five levels. Then, the clustering process is performed at each level, so that the size of the clusters varies in different levels. The proposed approach includes four phases: (1) Network leveling phase (2) Clustering phase (3) Key establishment phase (cluster key, pairwise key, and gateway key) (4) Rekeying phase. In the proposed method, a symmetric encryption algorithm called RC5 is used for producing keys. Our method contributes to the state of the art in wireless sensor networks by proposing a dynamic and multi-level key management approach. The use of a MAC-based authentication mechanism and the division of the network into different levels and clusters allows for efficient key establishment and rekeying, while minimizing energy consumption and communication overhead. Our proposed method is implemented using the NS2 simulator. Then, the simulation results are compared with SKWN and KMP methods. The results show that our proposed method outperforms SKWN and KMP in terms of average energy consumption, required memory, and communication overhead. However, its computational overhead is slightly more than SKWN and KMP.