Elsevier

Signal Processing

Volume 87, Issue 12, December 2007, Pages 2991-3009
Signal Processing

Diffuse: A topology building engine for wireless sensor networks

https://doi.org/10.1016/j.sigpro.2007.05.014Get rights and content

Abstract

In wireless sensor networks, multihop routing is commonly performed through a routing tree. Eventually, the routing tree needs to be rebuilt to accommodate failures, balance the energy consumption, or improve data aggregation. Most of the current solutions do not detect when the routing topology needs to be rebuilt. In this work, we propose an inference engine, called Diffuse, that uses information fusion techniques to detect when the routing topology needs to be rebuilt. Although the Diffuse applicability is ample, as a proof of concept, we use it to provide a fault-tolerant routing tree. Our solution is based on the data traffic that is pre-processed with a tunable moving average filter and translated into evidences that indicate failure probabilities. These evidences are combined using the Dempster–Shafer theory to determine when the topology needs to be rebuilt. We provide theoretical bounds for our proposed solution that is evaluated through simulation. Our experiments show that, in some cases, our solution can reduce the control traffic by a scale factor of 5.

Introduction

Wireless sensor networks (WSNs) [2] comprise a large number of nodes with sensing capability. The applicability of WSNs includes environmental, medical, industrial, and military applications. Commonly, WSNs are strongly computational and power limited. In addition, these networks demand self-organizing features to autonomously adapt themselves to eventual changes resulted from external interventions such as topological changes due to failures or node addition. An important task in WSNs consists in delivering the data gathered from the environment to a sink node for further processing and evaluation. Consequently, data routing is a fundamental task, which is commonly performed in a multihop fashion due to the radio range limitation and energy consumption constraints of a sensor node.

Tree topologies are frequently used to route data in a continuous flat sensor network [3], [4], [5], [6], [7]. Even the pioneer Directed Diffusion algorithm [8] provides a tree-like variant called One-Phase Pull Diffusion [7]. However, these solutions do not autonomously detect when the routing topology needs to be rebuilt. Although periodic [5] and user-triggered rebuilding are natural approaches to accommodate network changes, these strategies may lead to unnecessary reconstructions.

This work proposes the use of information fusion mechanisms to autonomously detect when the routing topology needs to be rebuilt. Briefly, information fusion deals with the combination of multiple sources to obtain improved information (e.g., cheaper, greater quality, or greater relevance). Information fusion is commonly used in detection and classification tasks in robotics and military applications [9]. Lately, it has been used in other applications such as intrusion [10] and Denial of Service (DoS) detection [11]. Within the WSNs domain, simple aggregation techniques (e.g., maximum, minimum, and average) have been used to reduce the overall data traffic to save energy [4], [8], [12].

The major contribution of this work is the improvement of routing algorithms by reducing unnecessary topology constructions, which is achieved by means of an inference engine called Diffuse. Diffuse is an ample solution that may be used to detect the need for topology reconstructions to pursue failure recovery, energy balance, and data aggregation improvements. As a proof of concept, we show how to implement Diffuse to allow the network to recover from routing failures avoiding unnecessary topology reconstructions. We also sketch how to extend Diffuse to other application domains. For the failure recovery evaluation, the simulation results show that, compared to the periodical rebuilding strategy, in some cases, our solution can reduce the control traffic by a scale factor of 5.

The rest of this paper is organized as follows. Section 2 presents the related work about tree topologies for multihop routing and failure detection for WSNs. Section 3 provides an overview of Diffuse, an inference engine, which is the building block we propose to detect the need for topology reconstructions. In Section 4, we show how Diffuse can be used to detect routing failures in WSNs, and present some theoretical results related to this scenario. Section 5 evaluates the rebuilding approaches through a set of simulations. In Section 6, we revisit the Diffuse applicability discussing how it can be extended to fulfill other applications. Finally, in Section 7, we present our conclusions and future work.

Section snippets

Related work

Among the current multihop routing strategies for flat sensor networks [3], [4], [5], [6], [7], [8], routing trees [3], [4], [5], [6] distinguish themselves from other approaches due to their simplicity and efficiency. Even the pioneer Directed Diffusion algorithm [8] provides a variant, called One-Phase Pull Diffusion [7], that implicitly builds a tree to route data towards the sink node.

Sohrabi et al. [3] present the Sequential Assignment Routing (SAR) algorithm to create multiple trees. Each

Diffuse: an inference engine

This section presents Diffuse, an inference engine that uses information fusion mechanisms to combine data and features to detect when the routing topology needs to be rebuilt. Diffuse may be schematically represented by the elements depicted in Fig. 1, as discussed below.

Diffuse for failure recovery

As discussed in Section 3.2, Diffuse can be used in different contexts. For the sake of exemplification, this section considers only the aspects that allow Diffuse to detect routing failures and rebuild the routing tree to reach disconnected nodes. In Section 6.2, we discuss again the applicability of Diffuse.

Simulation results

The simulation experiments performed in this section evaluate the use of Diffuse to detect routing failures and trigger topology reconstructions. The evaluation methodology and results are presented as follows.

Why Diffuse?

Since we use Diffuse to detect node failure, one might ask why use Dempster–Shafer if similar results could be obtained using a heartbeat approach. In this section, we try to elucidate when Diffuse is preferable to a failure detector based on heartbeats.

Conclusions and future work

This work presents an inference engine, called Diffuse, designed to detect when the routing topology needs to be rebuilt based on different goals, such as to recover from routing failures, improve data aggregation, and balance the energy consumption. For the sake of exemplification, we evaluate Diffuse aiming at recovering from routing failures. We provide theoretical results that show some properties of the periodic rebuilding, sink-centered, and source-centered approaches. These results show

References (21)

  • I.F. Akyildiz et al.

    Wireless sensor networks: a survey

    Comput. Networks

    (2002)
  • G. Provan

    The validity of Dempster–Shafer belief functions

    Internat. J. Approximate Reasoning

    (1992)
  • E. Nakamura, C. Figueiredo, A. Loureiro, Information fusion for data dissemination in self-organizing wireless sensor...
  • K. Sohrabi et al.

    Protocols for self-organization of a wireless sensor network

    IEEE Personal Comm.

    (2000)
  • B. Krishanamachari, D. Estrin, S. Wicker, The impact of data aggregation in wireless sensor networks, in: International...
  • C. Zhou, B. Krishnamachari, Localized topology generation mechanisms for self-configuring sensor networks, in:...
  • A. Woo, T. Tong, D. Culler, Taming the underlying challenges of reliable multihop routing in sensor networks, in:...
  • J. Heidemann, F. Silva, D. Estrin, Matching data dissemination algorithms to application requirements, in: Proceedings...
  • C. Intanagonwiwat, R. Govindan, D. Estrin, Directed diffusion: a scalable and robust communication paradigm for sensor...
  • R. Brooks et al.

    Multi-Sensor Fusion: Fundamentals and Applications

    (1998)
There are more references available in the full text version of this article.

Cited by (17)

  • Blind self-interference cancelation for amplify and forward relaying

    2017, 2017 10th Jordan International Electrical and Electronics Engineering Conference, JIEEEC 2017
  • Distributed fault detection based on HMM for wireless sensor networks

    2015, 2015 4th International Conference on Systems and Control, ICSC 2015
  • WSN implementation of DFD algorithm on SOFREL S550/WAVENIS unit

    2014, STA 2014 - 15th International Conference on Sciences and Techniques of Automatic Control and Computer Engineering
  • Analytical Modeling for reliability in cluster based Wireless Sensor Networks

    2014, Proceedings of 2014 9th IEEE International Conference on Computer Engineering and Systems, ICCES 2014
  • Decentralized fault detection in Wireless Sensor Network based on function error

    2013, 2013 10th International Multi-Conference on Systems, Signals and Devices, SSD 2013
View all citing articles on Scopus

This work extends a preliminary research [1]. The current version includes several new contributions: a generic framework for topology rebuilding; a more energy-efficient (in-network) algorithm; a new empirical evaluation, and several theoretical bounds. In addition, we show in this paper how our framework could be used to extend the current solution and provide a broader topology rebuilding system.

View full text