A Survey on Proactive, Active and Passive Fault Diagnosis Protocols for WSNs: Network Operation Perspective

Amjad Mehmood, Nabil Ali Alrajeh, Mithun Mukherjee, Salwani Abdullah, Houbing SongSensors 2018 · DOI

TL;DR

This paper surveys fault diagnosis protocols for Wireless Sensor Networks (WSNs) from a network operation perspective, categorizing them into proactive, active, and passive approaches. It highlights the critical need for fault diagnosis in WSNs due to their inherent limitations and challenging operating environments, offering a comprehensive overview of existing techniques, their advantages, limitations, and future research directions.

What Problem Does This Paper Address?

Wireless Sensor Networks (WSNs) are increasingly deployed in diverse and often harsh environments for applications ranging from environmental monitoring to scientific observation. These networks, composed of numerous resource-constrained nodes, are prone to failures due to factors like limited energy, environmental damage, and communication issues. While WSNs have seen extensive research in areas like reliability and robustness, the specific domain of in situ fault diagnosis for operational networks has received comparatively less attention. This gap is critical because undetected faults can lead to erroneous data, wasted energy, and a significantly shortened network lifespan, ultimately hindering the achievement of the network's objectives. The paper aims to bridge this gap by providing a structured survey of fault diagnosis protocols, focusing on their practical implications for network operations.

Methods

The paper categorizes existing fault diagnosis protocols into three main approaches:

  • Proactive Approaches: These involve embedding debugging agents on sensor nodes that periodically collect and transmit status information (e.g., residual energy, neighboring node status, battery health) to a sink node for analysis. While providing a wealth of information, they can generate significant traffic overhead and are often fragile under high traffic loads.
  • Active Approaches: In this paradigm, the sink node actively queries the network for specific information. Nodes within the query's proximity respond, but only if requested. This approach can also lead to substantial traffic overhead due to the transfer of control commands and status information.
  • Passive Approaches: Motivated by the limitations of proactive and active methods, passive approaches aim to consume fewer resources. They do not collect special data for diagnosis but instead perform diagnosis on normal data packets. These methods often leverage model-based techniques (like Artificial Neural Networks) or inference models to detect and diagnose faults by analyzing the consistency between expected and actual system behavior.

The survey reviews numerous specific schemes within each category, detailing their underlying techniques, advantages, and limitations from a network operation standpoint.

Datasets

Not discussed in the paper. The paper is a survey and does not present new experimental results or datasets.

Key Insights and Ideas

  • The Criticality of Fault Diagnosis in WSNs: The paper strongly emphasizes that fault diagnosis is not a luxury but a necessity for WSNs, directly impacting their lifespan, reliability, and ability to fulfill their intended functions, especially in challenging environments.
  • Trade-offs Across Approaches: Each fault diagnosis approach (proactive, active, passive) presents distinct trade-offs. Proactive methods offer rich data but incur high overhead. Active methods are query-driven but can also be bandwidth-intensive. Passive methods are resource-efficient but may rely on complex models or have limitations in dynamic scenarios.
  • The "Network Operation Perspective": A key contribution is the focus on how these protocols impact the day-to-day management and functioning of a WSN, considering factors like energy consumption, traffic overhead, and ease of deployment/maintenance.
  • The Evolution Towards Passive Methods: The survey highlights a trend towards passive approaches as a response to the resource constraints and operational challenges of WSNs, seeking more energy-efficient and less intrusive diagnosis mechanisms.
  • Intermittent Faults as a Major Challenge: The paper identifies intermittent faults as particularly difficult to monitor and detect due to their unpredictable nature, requiring specialized detection strategies.

Key Results

As a survey paper, it does not present quantitative experimental results. Instead, its key "results" are:

  • A comprehensive categorization and overview of proactive, active, and passive fault diagnosis protocols for WSNs.
  • A detailed analysis of the advantages and limitations of each scheme from a network operation perspective.
  • Identification of current research trends and open research challenges in the field.
  • A structured presentation of important terminology related to fault diagnosis.

Limitations and Future Work

The authors acknowledge several limitations and point towards future research directions:

  • Intelligent Techniques for Reduced Overhead: There's a strong demand for intelligent techniques that can minimize network overhead and bandwidth consumption, especially for online fault diagnosis in resource-constrained environments.
  • Efficient Load Balancing and Recovery: Protocols are needed for efficient load balancing and recovery of faulty sensor nodes, particularly for multimedia sensor nodes.
  • Energy-Efficient Protocols: Developing more energy-efficient protocols by analyzing end-to-end transmission delays, minimizing transmission range overlap, and reducing computational costs is crucial.
  • Safer and More Reliable Algorithms: The increasing demands of dynamic systems necessitate safer and more reliable algorithms.
  • Optimal Techniques for Node Identification: Optimal techniques are required to identify crashed (faulty) nodes, especially in scenarios where battery replacement is feasible or energy efficiency is paramount.
  • Low-Latency Fault Diagnosis: Given the deployment of WSNs in critical monitoring scenarios (e.g., gas leakage, fire), there's a need for fast, reliable, and fault-tolerant algorithms that can operate in query-based, periodic, or event-driven modes, ensuring data delivery even during emergencies.
  • Nonlinear and Uncertain Systems: Fault diagnosis techniques are needed for nonlinear, uncertain systems that rely on processing modeling.
  • Reliability and Scalability for QoS: Maintaining better Quality of Service (QoS) in large-scale WSNs is a significant challenge, requiring protocols that facilitate sensor consensus, interaction between different sensor types, and robustness in harsh environments.

Summary updated

BibTeX
@article{mehmood2018survey,
  author = {Amjad Mehmood and Nabil Ali Alrajeh and Mithun Mukherjee and Salwani Abdullah and Houbing Song},
  title = {{A Survey on Proactive, Active and Passive Fault Diagnosis Protocols for WSNs: Network Operation Perspective}},
  journal = {Sensors},
  year = {2018},
  doi = {10.3390/s18061787},
}
Figure