An Efficient Key Agreement Scheme for Wireless Sensor Networks Using Third Parties

TL;DR

This paper presents a secure and efficient key agreement scheme for wireless sensor networks that leverages third parties to establish pairwise keys, enhancing security, connectivity, and resource efficiency.

Contribution

The scheme introduces a novel third-party based key agreement method that ensures high local connectivity and security with minimal resource usage in sensor networks.

Findings

Achieves at least 99.42% local connectivity with 10% third-party nodes.

Ensures perfect resilience against node capture and impersonation attacks.

Requires less than 385 bits of memory per sensor node.

Abstract

This paper contributes to the challenging field of security for wireless sensor networks by introducing a key agreement scheme in which sensor nodes create secure radio connections with their neighbours depending on the aid of third parties. These third parties are responsible only for the pair-wise key establishment among sensor nodes, so they do not observe the physical phenomenon nor route data packets to other nodes. The proposed method is explained here with respect to four important issues: how secret shares are distributed, how local neighbours are discovered, how legitimate third parties are verified, and how secure channels are established. Moreover, the performance of the scheme is analyzed with regards to five metrics: local connectivity, resistance to node capture, memory usage, communication overhead, and computational burden. Our scheme not only secures the transmission channels of nodes but also guarantees high local connectivity of the sensor network, low usage of memory resources, perfect network resilience against node capture, and complete prevention against impersonation attacks. As it is demonstrated in this paper, using a number of third parties equals to 10% of the total number of sensor nodes in the area of interest, the proposed method can achieve at least 99.42% local connectivity with a very low usage of available storage resources (less than 385 bits on each sensor node).