A Lightweight Merkle Hash Tree-Based Framework for Black-Hole and Selective-Forwarding Attack Identification in Wireless Sensor Networks

Abstract

Wireless sensor networks (WSNs) forward data over multiple hops through resource-constrained and often unattended nodes, which makes black-hole and selective-forwarding attacks easy to mount and difficult to distinguish from ordinary packet loss. Existing detectors rely on neighbour overhearing, end-to-end acknowledgements or learned traffic models, and they either spend energy on promiscuous listening, localise attackers poorly, or provide no verifiable evidence. This paper proposes a Merkle hash tree (MHT) framework in which every forwarding node commits, once per epoch, to the sets of packets it received and forwarded using two constant-size authenticated roots (44 bytes in total). The sink audits these commitments with random leaf challenges, verifies the answers through logarithmic-size Merkle proofs against the next hop's committed root, and converts the resulting mismatch counts into node-level decisions with one-sided binomial tests that account for natural channel loss. In a simulation of 50 to 200 node networks with 5% to 25% compromised forwarders, half of which also falsified their own reports, the framework identified 100% of attackers with a false-positive rate of at most 0.18%, needed 3.2 epochs on average, and raised the packet delivery ratio from 44.8% to 82.4% after isolation. Control traffic is the main cost (57.2% of data bytes at the most responsive setting and 7.3% when epochs are lengthened and audits are sampled); this trade-off is quantified.