Although wireless sensor networks (WSNs) have been the object of research focus for the past two decades, fault diagnosis in these networks has received little attention. This is an essential requirement for wireless networks, especially in WSNs, because of their ad-hoc nature, deployment requirements and resource limitations. Therefore, in this paper we survey fault diagnosis from the perspective of network operations. To the best of our knowledge, this is the first survey from such a

more »

... ve. We survey the proactive, active and passive fault diagnosis schemes that have appeared in the literature to date, accenting their advantages and limitations of each scheme. In addition to illuminating the details of past efforts, this survey also reveals new research challenges and strengthens our understanding of the field of fault diagnosis. Fault diagnosis protocols Proactive Active Passive Model-based technique [68, 94] Classification framework [73] Inference model [78] Event-based routing [80] Debugging tool [26] Random walk approach [87] Nucleus management system [88] Sympathy-based approach [92] Mint-routing technique [91] Real-time monitoring [93] Analytical model [2] Lyapunov function [4] Probabilistic radio link [46] Blind forwarding [45] Top-k subgraph [47] UCR and two-state Markov model [64] Abstracted scan [89] Vector-based approach [65] DMTS [55] Probabilistic method [12, 53, 57, 59] Multipath scheduling [54] Figure 2. Fault diagnosis protocols.