Comprehensive Performance Evaluation of Wormhole Attack on RPL across Diverse IoT Network Environment

Abstract

The IPv6 Routing Protocol for Low-Power and Lossy Networks (RPL) is known to be sensitive to wormhole attacks, but there is a lack of systematic data on how tunnel multiplicity and node mobility interact to influence basic performance metrics. Quantified understanding of these consequences is critical for practitioners who must balance security costs with the tight energy and latency budgets of large-scale IoT installations in smart-city, industrial, and agricultural environments. We created a repeatable SimPy-based simulation environment that (i) embeds an arbitrary number of wormhole pairs in an RPL DAG, (ii) supports static, mobile, and hybrid topologies, and (iii) measures packet-delivery ratio (PDR), end-to-end latency, hop count, control overhead, energy consumption, and throughput. For 4 and 8 attacker pairs in each topology, three experimental modes were used: baseline, shortcut (forward tunnels), and black-hole (probabilistically dropping tunnels). In the static network, shortcut mode maintained PDR of around 1.0 while lowering mean latency by 18%, validating the tunnel's path-length advantage (hop count decreased from 2.80 to 2.13). When the same tunnels lost traffic, PDR fell to 0.96, 0.86, and 0.34 for four and eight attacker pairs, respectively, with no significant increase in control overhead. Mobility reduced the worst-case loss (mobile black-hole PDR recovered to 0.99 with four attackers), but it provided less relief as the number of tunnels rose. Hybrid cases in which attackers stayed static resulted in a 14% loss in delivery over eight tunnels. A small number of deliberately positioned wormholes can discreetly impair RPL dependability by an order of magnitude, but the forwarding variation is performance-neutral and hence impossible to detect. The findings highlight the importance of trust-zone placement or lightweight consistency checks when expanding RPL beyond low node counts, especially in static sensor fields where shortcut pathways are less obvious.