On Resilience of Distributed Flooding Algorithm to Stochastic Link Failures

Wireless sensor networks are vulnerable to various potential threats in numerous real-world applications. As recently identified by the scientific community, some faults can cause intensive damage to a system or even prevent the whole system from functioning properly. Therefore, the resilience of wireless sensor networks is an aspect that has gained importance over the last years. In this paper, we consider the scenario where the sensor-measured values are aggregated in highly defective wireless sensor networks using the distributed flooding algorithm operating in either the synchronous or asynchronous mode. More specifically, we assume that the communication among the nodes is affected by random link failures causing a complete loss of transmitted data. Our goal in this paper is to verify the applicability of the mentioned algorithm in both the synchronous and asynchronous mode, analyze how its precision is worsened, examine how the algorithm rate is affected, and identify which mode results in higher performance. The algorithm is tested in numerous randomly generated graphs (created by applying the Gilbert model) with various parameters and affected by link failures of different probabilities of occurrence. Also, the experimentally obtained results are compared to the performance of the algorithm in error-free scenarios.