An Efficient Distributed Trust Model for Wireless Sensor Networks

Trust models have been recently suggested as an effective security mechanism for Wireless Sensor Networks (WSNs).Considerable research has been done on modeling trust. However, most current research works only take communication behavior into account to calculate sensor nodes’ trust value, which is not enough for trust evaluation due to the widespread malicious attacks. In this paper, we propose an Efficient Distributed Trust Model (EDTM) for WSNs. First, according to the number of packets received by sensor nodes, direct trust and recommendation trust are selectively calculated. Then, communication trust, energy trust and data trust are considered during the calculation of direct trust. Furthermore, trust reliability and familiarity are defined to improve the accuracy of recommendation trust. The proposed EDTM can evaluate trustworthiness of sensor nodes more precisely and prevent the security breaches more effectively. Simulation results show that EDTM outperforms other similar models, e.g., NBBTE trust model. INTRODUCTION: WSNS are emerging technologies that have beenwidely used in many applications such as emergency response, healthcare monitoring, battlefield surveillance, habitat monitoring, traffic management, smart power grid, etc. However, the wireless and resource-constraint nature of a sensor network makes it an ideal medium for malicious attackers to intrude the system. Thus, providing security is extremely important for the safe application of WSNs.Various security mechanisms, e.g., cryptography, authentication,confidentiality, and message integrity, have been proposed to avoid security threats such as eavesdropping, message replay, and fabrication of messages. However, these approaches still suffer from many security vulnerabilities, such as node capture attacks and denial-of-service (DoS) attacks. The traditional security mechanisms can resist external attacks, but cannot solve internal attacks effectively which are caused by the captured nodes. To establish secure communications, we need to ensure that all communicating nodes are trusted. This highlights the fact that it is critical to establish a trust model allowing a sensor node to infer the trustworthiness of another node. From the literature on this topic, we can find that: 1) Inthe current research work, the assessment of trust values for sensor nodes is mainly based on the communication (successful and unsuccessful communications) point of view. In fact, just considering the communication behavior,we cannot decide whether a sensor node canbe trusted or not. Besides the communication behavior, other trust metrics such as the energy level should also be taken into account to calculate the trustworthiness ofsensor nodes. In addition, an efficient trust model should deal with uncertainty caused by noisy communication channels and unstable sensor nodes’ behaviors. 2) There are two common ways to establish trust in WSNs: calculating direct trust based on direct interactions and calculating indirect trust value based on recommendation from the third party. However, not all the third parties are trusty and not all the recommendations are reliable. Thus, a discriminate analysis about the third party and recommendation is essential. 3) Most existing studies only provide the trust assessment for neighbor nodes. However, in real applications, a sensor node sometimes needs to obtain the trust value of the nonneighbor nodes. For example, in some routing protocols or localization algorithms sensor nodes need the information of the two-hop neighbor nodes to establish the routing or localize themselves. Therefore, providing the trust assessment for non-neighbor nodes becomes very important.