A Reliable Information Fusion Algorithm for Reputation Based Wireless Sensor Networks

In wireless sensor networks (WSNs), cryptographic primitives alone cannot provide a sufficient solution to the secure information fusion problem, therefore reputation systems have been introduced into WSNs. In a cluster, each sensor node has a single reputation value that is evaluated by the other sensor nodes in the same cluster. In this paper, we propose a novel, reliable information fusion algorithm, called reputation-driven information fusion (RDIF). In this work, a clustering algorithm is employed to divide all of the sensor nodes into many clusters. Then, a reputation system is established for each cluster, and an information fusion algorithm driven by reputation values is performed by the cluster head. In addition to the sensor nodes’ reputation values, we also consider the values of the readings collected by the sensor nodes and eliminate the outliers before fusing information. The simulation results show that RDIF can improve the reliability and accuracy of fusion results significantly when some compromised nodes appear in the WSNs.