RTM: Realistic Weight-Based Reliable Trust Model for Large Scale WSNs

Trust estimation approaches in wireless sensor networks (WSNs) are mainly used to improve security, reliability, system efficiency, lifespan, decision-making, and collaboration (dependability) among sensor nodes. There are various existing trust estimation approaches incompetent to satisfy the fundamental requirements of WSN, like resource (power, energy, communication bandwidth) efficiency, long network lifetime, node availability, communication, and memory overheads. In this paper, we provide an efficient and accurate weight-based trust management approach that employs communication (direct, indirect) trust and data trust along with compressive sensing technique to alleviate several internal attacks like badmouthing, black-hole, and grey-hole attacks for clustered WSN. The core part of the proposed scheme is an efficient trust estimation function. The proposed trust estimation function computes the communication trust as well as data trust to analyze the cooperation level and data consistency respectively for better decision making. Although, existing schemes do not incorporate data trust during trust evaluation which leads to incorrect trust decisions. Moreover, the proposed reliable trust model which is known as RTM that provides adaptability and flexibility in terms of application requirements with reduced transmission overhead. RTM employs a simple, lightweight, and flexible approach for trust estimation. RTM uses a hybrid approach within clusters as well as with clusters to reduce communication overhead.RTM employs the concept of a directly (direct) trusted node and resolves the weight-assignment problem of existing weighted trust models of WSNs. Furthermore, this paper presents motivation and vision for WSNs as well as emphasizes the strength of the weighted approach by assigning suitable weights according to application requirements. The proposed trust model has been compared with the existing state-of-the-art trust models such as LDTS, LWTM, etc. Experimental results demonstrate excellent performance in terms of malicious behavior detection, prevention, the detection accuracy of selfish nodes, cooperation, as well as resource efficiency to protect WSN. RTM can detect 2.10%, 12.63%, 15.78% more malicious nodes than LWTM (Singh et al. in IETE J Res 63(3):297–308, 2017), ADCT (Talbi et al. in Telecommun Syst 65(4):605–619, 2017), LDTS (Li et al. in IEEE Trans Inf Forensics Secur 8(6):924–935, 2013) respectively in the network consisting of 500 nodes. Moreover, RTM can accurately detect 10%, and 12% more malicious nodes than LTS (Khan et al. in IEEE Access 7:58221–58240, 2019) and ADCT respectively in the presence of malicious nodes. Furthermore, RTM has 50%, 98% less inter-cluster communication overhead than LWTM, and ADCT in the presence of 100 clusters in the network.