Elsevier

Computer Communications

Volume 35, Issue 3, 1 February 2012, Pages 320-333
Computer Communications

Design of fault tolerant wireless sensor networks satisfying survivability and lifetime requirements

https://doi.org/10.1016/j.comcom.2011.10.006Get rights and content

Abstract

Sensor networks are deployed to accomplish certain specific missions over a period of time. It is essential that the network continues to operate, even if some of its nodes fail. It is also important that the network is able to support the mission for a minimum specified period of time. Hence, the design of a sensor network should not only provide some guarantees that all data from the sensor nodes are gathered at the base station, even in the presence of some faults, but should also allow the network to remain functional for a specified duration. This paper considers a two-tier, hierarchical sensor network architecture, where some relay nodes, provisioned with higher power and other capabilities, are used as cluster heads. Given a distribution of sensor nodes in a sensor network, finding the locations to place a minimum number of relay nodes such that, each sensor node is covered by at least one relay node, is known to be a computationally difficult problem. In addition, for successful and reliable data communication, the relay nodes network needs to be connected, as well as resilient to node failures. In this paper, a novel integrated Integer Linear Program (ILP) formulation is proposed, which, unlike existing techniques, not only finds a suitable placement strategy for the relay nodes, but also assigns the sensor nodes to the clusters and determines a load-balanced routing scheme. Therefore, in addition to the desired levels of fault tolerance for both the sensor nodes and the relay nodes, the proposed approach also meets specified performance guarantees with respect to network lifetime by limiting the maximum energy consumption of the relay nodes.

Introduction

A sensor network consists of tiny, low-powered and multi-functional sensor devices [2] and is able to perform complex tasks through the collaborative efforts of a large number of sensor nodes that are densely deployed within the sensing field [2]. In a sensor network, all data from the sensor nodes are collected at a special entity, called the base station. Each sensor node is operated by a battery, and usually, it is not feasible to replace or recharge this battery after deployment. Sensor nodes are equipped with a radio transceiver, and communicate using wireless media. The major source of power consumption in a sensor network is due to the wireless communication, which increases rapidly with distance between a transmitter and a receiver node. The lifetime (the period of useful operation) of a sensor network is considered over as soon as the battery power of the “critical node(s)” in the network is completely depleted [21], [30]. Therefore, tremendous attention has been paid in the literature to address the issue of energy conservation in the sensor nodes while gathering useful information, so that the lifetime of a sensor network can be extended as much as possible.

Recently, researchers have proposed the use of some special functionality nodes in sensor networks, henceforth referred to as relay nodes, to achieve various objectives, such as, to extend network lifetime, to achieve balanced data gathering, to reduce required transmission distance, to improve network connectivity and to enhance fault tolerance [10], [12], [14], [30]. These relay nodes can be provisioned with higher power and enhanced capabilities, as compared to the sensor nodes, and are ideally suited to serve as cluster heads in a hierarchical, two-tier sensor network [3], [16], [17], [18], [23], [24], [32]. In such networks, sensor nodes are partitioned into clusters and a relay node acts as a cluster head in each cluster. Each sensor node, lying in the lower-tier, belongs to a single cluster, and sends its data only to its cluster head. Therefore, each sensor node is relieved from the burden of routing and forwarding the data from other nodes. The relay nodes form the upper-tier of the network, and are responsible for collecting and forwarding data toward the base station. The lifetime of such a network (due to the depletion of battery power) is determined primarily by the power consumption of the relay nodes [18], [30]. Due to its numerous advantages including improved scalability and reliability [17], [18], [32], this paper focuses on such hierarchical, two-tiered model. It is assumed that sensor nodes are distributed appropriately in the sensing area, based on the needs of the specific application, and the locations of the sensor nodes are either known, or can be obtained, as in [17], [18], [32]. The design goal is to obtain the minimal number of relay nodes, along with their locations, such that (i) a certain given level of fault-tolerance, specified by the user based on the application requirement, is ensured, and (ii) the requirement for the minimum useful lifetime of the network, measured in term of power consumption of the relay nodes, is satisfied.

It is well known that sensor networks are prone to failures – due to node failure and/or communication failure. The failure of a relay node is more severe, as it not only results in data loss from all sensor nodes belonging to the cluster of the failed relay node, but may also disrupt information flow from other relay nodes, which are using the failed node for forwarding data towards the base station. Therefore, it is extremely important to have a placement strategy with sufficient degree of redundancy, so that each sensor node can send its data to multiple relay nodes, and there are several distinct paths from each relay node to the base station. The desired level of redundancy will depend on the intended application, and a generalized formulation should be capable of handling this. This paper presents an integer linear program (ILP) formulation, which incorporates fault tolerance by appropriately selecting the locations of a minimal number of relay nodes in the given networks, such that

  • Each sensor node can communicate with at least ks, (ks  1) relay node(s). This means that each sensor node can still transmit its data to at least one relay node, even if up to ks  1 relay node(s) (that the sensor node can communicate) fail. Henceforth, this property is referred to as the lower-tier network being ks-survivable.

  • Each relay node that uses a multi-hop path to the base station, is capable of communicating with at least kr, (kr  1) other suitable relay nodes, each having a valid path to the base station. This guarantees that each relay node has a viable path to the base station, even if up to kr  1 relay node(s) fail. Henceforth, this property is referred to as the upper-tier network being kr-survivable.

  • The maximum energy consumption of the relay nodes remains below a specified threshold.

The parameters ks and kr specify the desired levels of fault tolerance at the lower and the upper tiers of the network, respectively. The actual values of these parameters are determined by the application, and are given as inputs to the ILP.1 The objective is to achieve the desired levels of fault tolerances, in each tier of the network, with as few relay nodes as possible. A preliminary version of this approach was presented in [7]. However that study was limited to smaller networks, and it did not distinguish between the roles of primary and backup relay nodes.

Recently, some heuristics have been proposed in the literature for the special case where ks, kr = 1, and ks, kr = 2 [10], [32], [19], [27]. However, these are heuristic approaches that are not intended to handle arbitrary values of ks, kr. The formulation presented in this paper is the first generalized approach for the placement of relay nodes in a given sensor network that guarantees fault tolerance at both tiers, and ensures that the lower (upper) tier can tolerate up to ks  1 (kr  1) faults, for arbitrary values of ks and kr. The proposed formulation assumes that a set of potential positions, where the relay nodes can be placed, is computed separately based on the layout of the sensor nodes, and that the set includes sufficient redundant potential positions to meet the required levels of fault-tolerance.

Existing fault-tolerant placement algorithms [16], [32] typically do not take into account the energy consumption of the relay nodes, as this requires the knowledge of the routing scheme to be used. These approaches simply compute the minimum number, and the corresponding locations of the relay nodes such that the fault tolerance requirements are satisfied. However, for many applications, it may be essential to ensure a desired network lifetime. In order to achieve this, unlike current placement techniques, the proposed formulation jointly optimizes both placement and routing of relay nodes in two-tiered sensor networks, based on the pre-computed potential positions. The proposed approach not only designs a network that meets the level of fault tolerance at both tiers, but also finds a load-balanced routing schedule that ensures the energy consumption of each relay node does not exceed a user-specified limit.

The main contributions of this paper are as follows:

  • (i)

    An ILP formulation that jointly optimizes the placement and routing of relay nodes in a two-tiered sensor network such that, the network meets specified fault tolerance requirements, and also meets the constraints for maximum energy consumption by the relay nodes. The proposed approach can handle different data rates among the sensor nodes.

  • (ii)

    Two heuristic approaches, a grid based approach and an intersection based approach, for determining the potential positions of the relay nodes.

  • (iii)

    Simulation results to demonstrate that the proposed joint optimization approach can meet both survivability and lifetime requirements, while the traditional two-step approaches are typically unable to achieve this.

The remainder of this paper is organized as follows. In Section 2, some background information is reviewed. Section 3 discusses the network model and presents the ILP formulations. Heuristics to compute the potential positions of relay nodes are discussed in Section 4. The simulation results are presented in Section 5, and the conclusion is given in Section 6.

Section snippets

Background review

Data communication, from relay nodes to the base station, may be either single-hop (where each relay node receives data only from its own cluster, and sends this data directly to the base station) [20], [21] or multi-hop (where relay nodes form an upper-tier network among themselves, and each relay node, in addition to forwarding the data it receives from its own cluster, also forwards the data it receives from other relay nodes, towards the base station, using multi-hop paths) [23], [24], [26]

Formulation for fault tolerant placement and routing

In this section, the ILP formulation that jointly considers the survivability and the lifetime requirement is discussed.

Heuristic for finding the potential locations of relay nodes

In the previous section, an ILP formulation is presented that optimally selects the positions of the relay nodes from a set of potential positions, and determines a routing schedule that meets certain energy requirements for some given values for ks and kr. Experimental results (discussed in Section 5) demonstrate that adding a few properly placed relay nodes can significantly extend the network lifetime. In this context it is extremely important that the set of “potential” relay node positions

Simulation results

In the simulation results discussed in this section, it is assumed that the communication range of sensor nodes is rmax = 40 m, and the range of a relay node is dmax = 200 m, as in [32], unless stated otherwise. It is also assumed that

  • (1)

    the communication energy consumption is based on the first order radio model, described in Section 2 and

  • (2)

    the values for the constants are the same as in [21], so that:

    • (a)

      α1 = α2 = 50 nJ/bit,

    • (b)

      β = 100 pJ/bit/m2 and

    • (c)

      the path-loss exponent, q = 2.

  • (3)

    the initial energy of each relay node was 5J

Conclusions

In two-tiered, cluster based sensor networks using relay nodes as cluster heads, the relay nodes placement problem and the routing problem of the upper-tier network, are conventionally solved separately. In this paper, an ILP formulation that solves these problems jointly is proposed. This approach requires a set of potential locations of the relay nodes. Two heuristic approaches to determine the set of such potential locations are also proposed. The formulation takes the set of potential

Acknowledgment

A. Bari and J. Jiang would like to acknowledge the financial support for this work from the Natural Sciences and Engineering Research Council of Canada (NSERC); the Ministry of Research and Innovation (MRI), Ontario, Canada; the University Network of Excellence in Nuclear Engineering (UNENE), and ISTP Canada.

A. Jaekel would like to acknowledge the financial support from the NSERC for this work.

References (42)

  • M. Borghini et al.

    Optimal data delivery in wireless sensor networks in the energy and latency domains

    WICON

    (2005)
  • X. Cheng et al.

    Relay sensor placement in wireless sensor networks

    IEEE Transactions on Computers

    (2001)
  • S. Coleri, P. Varaiya, Optimal placement of relay nodes in sensor networks, in: the Proceedings of ICC,...
  • K. Dasgupta, M. Kukreja, K. Kalpaki, Topology-aware placement and role assignment for energy-efficient information...
  • S.C. Ergen, P. Varaiya, Optimal placement of relay nodes for energy efficiency in sensor networks, in: ICC,...
  • E. Falck, P. Floren, P. Kaski, J. Kohonen, P. Orponen, Balanced data gathering in energy-constrained sensor networks,...
  • G. Gupta, M. Younis, Load-balanced clustering of wireless sensor networks, in: IEEE ICC 03, vol. 3, 2003, pp....
  • G. Gupta, M. Younis, Fault-tolerant clustering of wireless sensor networks, in: Proceedings of IEEE WCNC, 2003, pp....
  • G. Gupta, M. Younis, Performance evaluation of load-balanced clustering of wireless sensor networks, in: 10th...
  • B. Hao, J. Tang, G. Xue, Fault-tolerant relay node placement in wireless sensor networks: formulation and...
  • W. Heinzelman, Application-specific Protocol Architectures for Wireless Networks. Ph.D. thesis, Massachusetts Institute...
  • Cited by (79)

    • An adaptive cellular automata scheme for diagnosis of fault tolerance and connectivity preserving in wireless sensor networks

      2018, Alexandria Engineering Journal
      Citation Excerpt :

      Table 1 compares between related cluster-based fault tolerant techniques. More formal description, as well as a discussion of these algorithms, can be found in [16–24]. In the previous cluster-based FT studies, no attention is given to QoS concepts like coverage and connectivity [5].

    • Energy Efficient and Node Density Based Dual Base Station Deployment for Large Scale Wireless Sensor Network

      2023, International Journal of Intelligent Systems and Applications in Engineering
    • Efficient Data Flow Graph Modeling Using Free Poisson Law for Fault-Tolerant Routing in Internet of Things

      2023, Lecture Notes on Data Engineering and Communications Technologies
    • Distributed topology fault-tolerant time-varying formation control based on persistent topology optimization and reconfiguration

      2022, Proceedings - 2022 37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022
    View all citing articles on Scopus
    View full text