Cascading Failures Analysis of Wireless Sensor Networks With Varying Routing Schemes

In practical wireless sensor networks (WSNs), routing schemes determine the message-forwarding direction of sensor nodes, which have a significant impact on the cascading process of WSNs. However, the existing cascading failure models fail to consider the impact of routing schemes. Therefore, in this work, we propose an actual cascading failure model for WSNs considering routing factors. In this model, a sink-oriented load metric is proposed to characterize the actual load distribution of WSNs, and four load redistribution mechanisms (i.e., latency-sensitive global routing, congestion-sensitive global routing, capacity-sensitive local routing and latency-sensitive local routing) are proposed to mimic the most commonly used routing schemes. Through extensive simulations, we have obtained some meaningful results: 1) there is a critical threshold for node capacity. When the node capacity is greater than the threshold, the lifting effects of capacity expansion are saturated; 2) there is a cascade interval for the attack range. Cascading failures occur only if the attack range is within this interval; 3) among the four routing schemes, congestion-sensitive global routing has the strongest cascading robustness; 4) flat-structured topology is the most robust, followed by clustering topology, and finally scale-free topology. The obtained results can provide theoretical support for improving the cascading robustness of WSNs.