Elsevier

Computer Communications

Volume 155, 1 April 2020, Pages 205-226
Computer Communications

Review
Fault management frameworks in wireless sensor networks: A survey

https://doi.org/10.1016/j.comcom.2020.03.011Get rights and content

Highlights

  • Classification of faults in a sensor network based on faults behavior.

  • Classification of fault management components and types in 3 steps.

  • Classification of fault management frameworks as general classes and subclasses.

  • Describing current frameworks and representing major steps in fault management.

  • Evaluating and analyzing of represented frameworks based on their main challenges.

Abstract

Wireless sensor networks (WSNs) are composed of a set of sensor nodes, widely spread out across the area and gather information. Some sensor nodes may develop a fault due to various reasons such as environmental effects. Since the occurrence of faults should not affect the desired function and functionality of network, fault management is vital to improve fault tolerance. Here, we classify faults based on their behavior, durability, and the components of network. Then the components of fault management and the challenges of fault management frameworks are represented. Also, other frameworks are analyzed to define their major challenges including energy consumption, fault detection accuracy, delay, the scalability and overhead of network. Other papers have classified fault management frameworks into centralized, distributed, and hierarchical, but we suggest a new classification based on the performance of management of each framework and the number of involved nodes. Next, frameworks have been analyzed and evaluated according to their major challenges. The analyses have provided a chance to offer more accurate and effective fault management frameworks with minimum energy consumption, delay, and overhead.

Introduction

WSNs are self-organized networks, including several sensor systems with limited capability of processing, storing, communication, and limited battery capacity. Now, sensor networks are increasingly used in applied areas including medicine, supervising environment, and forests [1].

The main applications of fault management in WSNs could be classified into 3 groups: constantly accessible systems, real-time systems, and accurate systems. Thus fault management 1and increased fault tolerance is necessary for these networks. It is expected that sensor nodes operate independently in various areas for a long period of time and may not be accessible for maintenance and battery replacement due to their physical situation. Faults may occur in different components of network and are of different types based on durability and effects on the performance of nodes [2]. Therefore, fault management is vital to guarantee the quality of services and performance of networks. A set of functions or applications, specifically designed for fault tolerance, is called fault management framework and is an inseparable part of network management system [3]. The fault management of WSNs is an important component of network management, including detection, diagnosis, and recovery of faults. Proper implementation of fault management can result in optimal network management and decreased network fault, leading to increased fault tolerance. In addition, certain solutions are required to reduce energy consumption and maximize network lifetime [4].

In recent years, authors have designed several fault management frameworks for detection and recovery of faults. Here, we have searched for research studies on fault management frameworks in WSNs through the keywords represented in Table 1. The articles are searched for in Google Scholar and academic databases such as Springer Link, Science Direct, IEEE explore, ACM Digital Library, Wiley Interscience, and Taylor & Francis Online.

On finding 2258 articles, we chose 663 papers based on their titles. Reviewing the abstracts and evaluating their relation, we selected 103 articles. Then 54 papers on fault management frameworks were selected to be investigated in this article, including 10 review articles, which are represented in Fig. 1. We scrutinized 34 frameworks suggested between 2003 and 2018.

Most of review articles do not include evaluation of frameworks and usually focus on classification of faults or general review of frameworks. Thus we seek to provide a comprehensive review and proper analysis. The main objectives of our article are:

  • Classification of faults in a sensor network based on faults behavior and durability, components of network, and the area affected by faults

  • Classification of fault management components and types in 3 steps: detection, diagnosis, and recovery. Also, suggesting a new classification for fault recovery methods based on their performance

  • Classification of fault management frameworks as general classes and subclasses based on their performance

  • Describing current frameworks and representing major steps in fault detection, diagnosis, and recovery

  • Evaluating and analyzing represented frameworks based on their main challenges and providing a chance for suggesting more effective frameworks in sensor networks

Here, fault management frameworks are classified into 4 groups: cartelized, distributed, hierarchical, and hybrid. In centralized frameworks, a sensor node logically and geographically manages the whole network. The node (central node) is usually provided with unlimited resources and is able to manage faults. By analyzing the steps and performance of fault management frameworks, we classify them into 3 classes based on the mechanisms applied in the center: database-based, machine learning-based, and time-based. To rectify the faults of centralized frameworks, distributed frameworks were introduced. Distributed fault management frameworks distribute fault management tasks throughout the network. The frameworks are classified into 2 groups based on the number of nodes involved in fault management: agent-based and neighbor cooperation-based. Hierarchical fault management frameworks, a combination of centralized and distributed methods, were introduced to improve distributed ones. Evaluating the performance of fault management frameworks, we classified them into 4 groups: database-based, time-based, neighbor cooperation-based, self-management-based. Hybrid framework is developed through combining distributed and hierarchical frameworks. Current frameworks are investigated based on their challenges including energy consumption, fault detection accuracy, scalability, and traffic load.

The main purpose of each section of the paper is to answer the key questions that are discussed below.

  • 1.

    What are the main faults affecting the WSNs that need to be managed? (Section 2)

  • 2.

    What are the fault management steps in the WSNs and the methods of each step in the sensor network? (Section 3)

  • 3.

    How can a careful categorization be presented for fault management frameworks? (Section 4)

  • 4.

    What are the most important challenges of fault management frameworks that need to be evaluated? (Section 4)

  • 5.

    How can fault management frameworks in each class be compared? (Sections 5, 6, 7, 8)

  • 6.

    Given the challenges of WSNs, how can we effectively evaluate the fault management frameworks in a way that provides a better framework in the future? (Section 9)

The paper is organized as follows. In the following, fault classifications are addressed. Then the fundamentals of fault management are represented in Section 3. Section 4 includes the proposed classification of frameworks and challenges. In Section 5, Section 6, Section 7 and Section 8, respectively the centralized, distributed, hierarchical and hybrid frameworks are included. Section 9 includes discussion and evaluation of frameworks. The final section includes conclusions and future work.

Section snippets

Fault classification for WSN

The main characteristics of faults in WSNs are the cause and origin of faults. The causes of faults are classified into 4 groups: improper definition of network specifications, improper implementation of network, failure of system components, and environmental elements. Improper definition of network specifications by designers and programmers results in faults [5]. On designing, improper implementation of network leads to faults. Even if designing and implementing are faultless, some

Fundamentals of fault management framework in WSNs

In this section, we describe the fundamentals of fault management framework in WSN. Fault tolerance refers to a capability of system that allows it to deal with faults and maintain its optimal activity [1]. One of the common approaches to increase fault tolerance is fault management. Fault management in WSNs includes 3 steps: detecting faults, positioning faulty nodes, and recovering faults. In the following, the three steps of the fault management framework and their evaluation are described.

Proposed classification of fault management frameworks and challenges in WSNs

In order to clarify the presentation of fault management frameworks, we define a new classification in this section. In the proposed classification, fault management frameworks are divided into four groups: centralized, distributed, hierarchical, and hybrid. Each group includes some subgroups. The proposed classification of fault management frameworks is shown in Fig. 4.

Centralized frameworks, which manage faults in the center or sink, are classified into 4 groups based on the mechanism:

Centralized fault management frameworks

In this section, centralized fault management frameworks are outlined. Fault management has a central manager, controlling and monitoring network. A manager is a tool or process used by BS to manage nodes. Making changes to the setting of nodes is merely possible through connecting to manager. On detecting a fault by the manager, notification process begins. Notification is a process applied by the manager to alarm BS to start fault management process.

In centralized fault management systems, a

Distributed fault management frameworks

In this section, distributed fault management frameworks are outlined. These frameworks employ a multiple management stations in whole network instead of a central control. In fact, the main purpose of providing the frameworks is to avoid some problems of centralized frameworks: overhead and scalability. In distributed frameworks, each manager controls a network subset and can directly communicate with other management stations to perform management functions. Distributed frameworks lead sensor

Hierarchical fault management frameworks

In this section, the hierarchical fault management frameworks are outlined. These frameworks are a combination of centralized and distributed methods. Generally, hierarchical frameworks represent an extra level of nodes for management, a method which makes distribution of control simpler. The frameworks are suggested to deal with delay and complexity of distributed structure. Hierarchical management structures are based on clusters and cells.

In the frameworks based on clustering, sensor nodes

Hybrid fault management framework

In [54], a hybrid framework, a combination of hierarchical and distributed frameworks, is suggested. The framework is called LPS-FMP and is composed of management station, management system, agent system, gateway system, and normal sensor nodes. In this framework, the information of faults is gathered through 2 methods: (1) agent uploads information (2) management sends out queries. To detect faults, 3 components are involved: fault detection system, gateway detection system, and Client

Discussion and evaluation

According to the challenges of designing fault management frameworks, represented in Section 4.1, the aforementioned studies have been evaluated and analyzed. The challenges include energy consumption, delay, scalability, and overhead. In the following, the evaluation of the current frameworks is represented.

Conclusions

As nodes are prone to faults and failure in WSNs, fault management frameworks are required. Now, several fault management frameworks have been suggested and their classification and analysis allows understanding fault management approaches. Here, fault management frameworks in WSNs have been addressed. Before analyzing frameworks, we need to classify faults in sensor networks as each framework detects and recovers one type of faults. In this study, we classified the frameworks into 4 groups:

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (89)

  • MitraS. et al.

    A review of fault management system in wireless sensor network

  • KhanM.Z.

    Fault management in wireless sensor networks

    Comput. Sci. Telecommun.

    (2013)
  • KeerthanaS. et al.

    A survey on fault node detection and recovery mechanisms in wireless sensor network

  • ZhangY. et al.

    A framework and classification for fault detection approaches in wireless sensor networks with an energy efficiency perspective

    Int. J. Distrib. Sens. Netw.

    (2015)
  • ZhangZ.

    A survey on fault diagnosis in wireless sensor networks

    IEEE Access

    (2018)
  • KarlH. et al.

    Protocols and Architectures for Wireless Sensor Networks

    (2007)
  • ShyamaM. et al.

    Fault-tolerant techniques for wireless sensor network—A comprehensive survey

  • DiazS. et al.

    A review on self-healing and self-organizing techniques for wireless sensor networks

    J. Circuits Syst. Comput.

    (2019)
  • RajiK.A. et al.

    A survey of different techniques for energy-efficient, reliability and fault tolerant in wireless sensor networks

    Int. J. Wirel. Commun. Mob. Comput.

    (2019)
  • De SouzaL.M.S. et al.

    A survey on fault tolerance in wireless sensor networks

    Int. Bericht. Fak. Inform. Univ. Karlsruhe

    (2007)
  • ParadisL. et al.

    A survey of fault management in wireless sensor networks

    J. Netw. Syst. Manage.

    (2007)
  • ZHUY.-j.

    Status and analysis of fault diagnosis in wireless sensor networks

    DEStech Trans. Comput. Sci. Eng.

    (2019)
  • PandaM. et al.

    Distributed online fault diagnosis in wireless sensor networks

  • ShihH.-C.

    Fault node recovery algorithm for a wireless sensor network

    IEEE Sens. J.

    (2013)
  • CuriacD.-I.

    Redundancy and its applications in wireless sensor networks: A survey

    WSEAS Trans. Comput.

    (2009)
  • łgorzata SteinderM. et al.

    A survey of fault localization techniques in computer networks

    Sci. Comput. Program.

    (2004)
  • VihmanL. et al.

    Overview of fault tolerant techniques in underwater sensor networks

    (2019)
  • MallickC. et al.

    Challenges and design goals of wireless sensor networks: a sate-of-the-art review

    Int. J. Comput. Appl.

    (2018)
  • AsimM. et al.

    A fault management architecture for wireless sensor network

  • MuthukarpagamS. et al.

    Design issues, topology issues, quality of service support for wireless sensor networks: Survey and research challenges

    Int. J. Comput. Appl.

    (2010)
  • PengY. et al.

    Survey of fault management framework in wireless sensor networks

    J. Electr. Meas. Instrum.

    (2009)
  • R.V. Kshirsagar, B. Jirapure, A survey on fault detection and fault tolerance in wireless sensor networks, in: IJCA...
  • DhulipalaV.S. et al.

    Energy and fault aware management framework for wireless sensor network

  • MahapatroA. et al.

    Choice of detection parameters on fault detection in wireless sensor networks: A multiobjective optimization approach

    Wirel. Pers. Commun.

    (2014)
  • SokulluR. et al.

    Fault management for smart wireless sensor networks

  • KimY.J. et al.

    HDF: Hybrid debugging framework for distributed network environments

    ETRI J.

    (2017)
  • TolleG. et al.

    Design of an application-cooperative management system for wireless sensor networks

  • H. Song, et al. UPnP-based sensor network management architecture, in: Proc. International Conference on Mobile...
  • YiW.-Y. et al.

    A modular plug-and-play sensor system for urban air pollution monitoring: design, implementation and evaluation

    Sensors

    (2018)
  • TuronM.

    Mote-view: A sensor network monitoring and management tool

  • KumarD.P. et al.

    Machine learning algorithms for wireless sensor networks: A survey

    Inf. Fusion

    (2019)
  • AlsheikhM.A.

    Machine learning in wireless sensor networks: Algorithms, strategies, and applications

    IEEE Commun. Surv. Tutor.

    (2014)
  • de SouzaL.S.

    FT-CoWiseNets: A fault tolerance framework for wireless sensor networks

  • SalemO.

    Sensor fault and patient anomaly detection and classification in medical wireless sensor networks

  • Cited by (45)

    • Recent trends in clustering algorithms for wireless sensor networks: A comprehensive review

      2022, Computer Communications
      Citation Excerpt :

      The centralized approach reduces energy overhead, but it is not applicable in large-scale networks. More details about fault tolerance WSNs is presented in [344,358,359]. The clustering technique is widely used in designing routing protocols for WSNs because it offers many advantages such as scalability, efficient data aggregation, fault tolerance, latency reduction, robustness, and reduced energy consumption.

    View all citing articles on Scopus
    View full text