Elsevier

Ad Hoc Networks

Volume 4, Issue 3, May 2006, Pages 416-430
Ad Hoc Networks

Locating hot nodes and data routing for efficient decision fusion in sensor networks

https://doi.org/10.1016/j.adhoc.2004.11.001Get rights and content

Abstract

This paper presents a fundamentally new approach to integrating local decisions from various nodes and efficiently routing data in sensor networks. By classifying the nodes in the sensor field as “hot” or “cold” in accordance with whether or not they sense the target, we are able to concentrate on a smaller set of nodes and gear the routing of data to and from the sink to a fraction of the nodes that exist in the network. The introduction of this intermediary step is fundamentally new and allows for efficient and meaningful fusion and routing. This is made possible through the use of a novel Markov Random Field (MRF) approach, which, to the best of our knowledge, has never been applied to sensor networks, in combination with Maximum A Posteriori Probability (MAP) stochastic relaxation tools to flag out the “hot” nodes in the network, and to optimally combine their data and decisions towards an integrated and collaborative global decision fusion. This global decision supersedes all local decisions, and provides the basis for efficient use of the sensed data. Because of the MRF local nature, nodes need not see or interact with other nodes in the sensor network beyond their immediate neighborhood, which can either be defined in terms of distance between nodes or communication connectivity, hence adding to the flexibility of dealing with irregular and varying sensor topologies, and also minimizing node power usage and providing for easy scalability. The routing of the “hot” nodes’ data is confined to a cone of nodes and power constraints are taken into account. We also use the found location of the centroid of the hot nodes over time to track the movement of the target(s). This is achieved by using the segmentation at time t as an initial state in the stochastic MAP relaxation at time t + Δt.

Introduction

The deployment of large scale sensor networks is becoming imminent, and so is the need to resolve the challenges related to this type of network. Sensor networks differ from traditional wireless ad-hoc networks in several aspects, the most important being the number of nodes which can be of several orders of magnitude higher than in ad hoc networks, the node density, the fact that sensor nodes are prone to failure, the dynamic topology of a sensor network, which changes very frequently, and the fact that sensor nodes are limited in power, computational capacity, and memory [1].

This paper concentrates on contributions to the networking layer of sensor networks. The following considerations are critical to its design: data fusion (to combine the information collected from several sensors), location awareness for scalability, power efficiency (due to the limited lifetime of a sensor node), and data centric networks (query for the data). Moreover, the proposed mechanisms have to cope with a fast changing topology, as a function of node failure and/or mobility.

An excellent survey on sensor networks is presented in [1]. A survey on routing protocols is given in [2]. According to that survey, routing protocols can be classified into data centric, hierarchical and location based.

Flooding [3], Gossiping [4], SPIN [3], Directed Diffusion [5], Energy Aware Routing [6], and Rumor Routing [7] are examples of data centric protocols. In such protocols, the nodes are queried based on some attributes of the data (the data is named), and therefore no specialized nodes are needed and the overhead of the communication is minimized. However, the naming schemes are usually application dependent and further research needs to be carried out for an efficient naming scheme capable of handling complex queries.

LEACH [8], TEEN [9], Energy Aware Routing for Cluster-Based Sensor Networks [10], and self-organizing protocol [11] are examples of hierarchical protocols. In such protocols, nodes are grouped into clusters and a “cluster-head” is selected, sometimes randomly and sometimes as a specialized node with less power constraints. The cluster-head is responsible for aggregating the cluster data and delivering it to the sink. Further research needs to be performed in order to minimize the overhead when forming clusters, to optimize cluster-head communication, and to integrate data and decision and fuse them with cluster formation.

MECN [12], SMECN [13], GAF [14], and GEAR [15] are location based protocols. Although it has been shown in [16] that routing protocols that do not use geographical location information are not scalable (e.g., traditional ad-hoc protocols such as DSR and AODV), the number of proposed energy-aware location based protocols in still quite small. Research challenges include the efficient use of the location information in an energy efficient routing mechanism, and the interaction with data aggregation and fusion mechanisms.

This paper examines a new approach to sensor networks in which the integration of a robust and scalable data and decision fusion mechanism coupled with an energy efficient location-based routing algorithm is presented. We take into account the limited power, computational capacity, and memory of sensor nodes, the large number and or density of nodes in a sensor field (scalability), and also the fact that the topology may change very frequently due to node failure. In our approach, nodes need not know of or interact with other nodes in the field beyond their immediate neighbors (which also provides for energy savings). Also, the possibility of incomplete data is also considered in the model and an accurate estimation method is proposed. Our decision and data fusion approach identifies “hot” nodes which are in possession of important data that needs to be routed back to the sink.

Section snippets

Finding hot nodes

Let H0 be the hypothesis that there is no target and H1 be the alternative hypothesis that there is a target at given location in the sensor network, with either a regular or irregular topology as shown in Fig. 1.

Let Yi be the sensed data or some function (a sufficient statistics) of the data at node i, and let Xi denote the state of node i with Xi = 1 means that the node senses the target and Xi = 0 means it does not. The statistics of Y is either an implicit or explicit function of the target

Combining information from hot nodes

The result of the labeling process is the partitioning of the sensor network into “hot” and “cold” nodes. A global decision can be achieved by combining all the local decisions associated with the local “hot” nodes into a global Newman Pearson (NP) decision fusion based on the likelihood ratio statistics. The NP has the highest probability of detection for a fixed false alarm rate amongst all other decision rules.

The local Newman Pearson decision rulep(Yi|H1)>H1p(Yi|H0)H0<ζi,reduces toXi=1iffTi>

Routing data from hot nodes

Once the “hot” nodes are identified, the problem of routing their data back to the sink is triggered. Note that the “hot” nodes might be located in more than one area/section in the sensor field (see Fig. 6) and, most likely, such an area/section will contain a number of “hot” nodes.

In our model, the sensor nodes have a small transmission range, while the sink is equipped with a much more powerful radio device and also a much more powerful battery. A sensor node that is not in the proximity of

Experiments

We illustrate the framework by considering the following simulation example. We consider a network consisting of about 26,000 nodes. At each node s at location ds (relative to the sink in the network) there is a sensor, for example an ultrasound sensor that records the backscatter echo of a possible target at location t in its vicinity. The backscatter echo at node s is assumed to be modeled asYs=Es+ns,where the echo Es follows a Gaussian function which tapers off away from the target, with the

Conclusions and discussion

We developed a probabilistic Bayesian Markov Random Field (MRF) approach to sensor networks where MAP tools are used to find “hot” nodes in the sensor field. This intermediary step that is fundamentally new, and allows for efficient and meaningful fusion and routing, for optimally integrating and fusing the data and/or decisions at the hot nodes to test various hypotheses ranging from target recognition and identification, to monitoring and tracking, to tracking of traffic jams and accidents,

Fernand S. Cohen received his B.Sc. degree in Physics from the American University in Cairo in 1978, and M.Sc. and Ph.D. degrees in Electrical Engineering from Brown University, Providence, RI, in 1980 and 1983, respectively. He joined the Department of Electrical Engineering at the University of Rhode Island in 1983 as an Assistant Professor. In 1984 he joined the Robotics Research Center, University of Rhode Island and was responsible for the Vision Research in the center from 1986 to 1987.

References (24)

  • I.F. Akyildiz et al.

    Wireless sensor networks: a survey

    Comput. Networks

    (2002)
  • K. Akkaya, M. Younis, A survey on routing protocols for wireless sensor networks, Ad Hoc Networks, in press;...
  • W. Heinzelman, J. Kulik, H. Balakrishnan, Adaptive protocols for information dissemination in wireless sensor networks,...
  • S. Hedetniemi et al.

    A survey of gossiping and broadcasting in communication networks

    Networks

    (1988)
  • C. Intanagonwiwat, R. Govindan, D. Estrin, Directed diffusion: a scalable and robust communication paradigm for sensor...
  • R. Shah, J. Rabaey, Energy aware routing for low energy ad hoc sensor networks, in: Proceedings of the IEEE Wireless...
  • D. Braginsky, D. Estrin, Rumor routing algorithm for sensor networks, in: Proceedings of the First Workshop on Sensor...
  • W. Heinzelman, A. Chandrakasan, H. Balakrishnan, Energy-efficient communication protocols for wireless microsensor...
  • A. Manjeshwar, D.P. Agrawal, TEEN: a protocol for enhanced efficiency in wireless sensor networks, in: Proceedings of...
  • M. Younis, M. Youssef, K. Arisha, Energy-aware routing in cluster-based sensor networks, in: Proceedings of the 10th...
  • L. Subramanian, R.H. Katz, An architecture for building self configurable systems, in: Proceedings of IEEE/ACM Workshop...
  • V. Rodoplu et al.

    Minimum energy mobile wireless networks

    IEEE J. Selected Areas Commun.

    (1999)
  • Fernand S. Cohen received his B.Sc. degree in Physics from the American University in Cairo in 1978, and M.Sc. and Ph.D. degrees in Electrical Engineering from Brown University, Providence, RI, in 1980 and 1983, respectively. He joined the Department of Electrical Engineering at the University of Rhode Island in 1983 as an Assistant Professor. In 1984 he joined the Robotics Research Center, University of Rhode Island and was responsible for the Vision Research in the center from 1986 to 1987. In 1985 he was the recipient of a Research Excellence Award from the College of Engineering, University of Rhode Island. In 1986 he was invited by the French government (mission scientifique) to tour research laboratories and universities. In 1987 he joined the Department of Electrical and Computer Engineering at Drexel University as a named Chair Associate Professor (George Beggs). He is currently Professor of Electrical and Computer Engineering and is affiliated with the School of Biomedical Engineering, Science and Health Systems, and serves as Director of the Imaging and Computer Vision Center (ICVC). In the summer 1994 he was invited as a visiting Professor by the National Institute of Research in Information and Automation (INRIA) in Sophia Antipolis, France. In 2003 he was the recipient of the Thomas W. Moore Award for Excellence and Innovation in teaching from the Electrical and Computer Engineering Department at Drexel University. His research interests include pattern recognition, signal processing, computer vision, medical image processing, and applied stochastic processes.

    Jaudelice C. de Oliveira received her B.S.E.E. degree from Universidade Federal do Ceara (UFC), Ceara, Brazil, in December 1995. She received her M.S.E.E. degree from Universidade Estadual de Campinas (UNICAMP), Sao Paulo, Brazil, in February 1998, and her Ph.D. degree in Electrical and Computer Engineering from the Georgia Institute of Technology in May 2003. She joined Drexel University in July of 2003 as an Assistant Professor. Her research interests include the development of new protocols and policies to support fine grained quality of service provisioning in the future Internet, researching and developing traffic engineering strategies for MultiProtocol Label Switching (MPLS) networks, the design of solutions for managing heterogeneous and large computer networks, and energy aware medium access and routing protocols for ad hoc networks.

    Ezgi Taslidere received her B.S. and M.S. degrees in Electrical and Electronics Engineering from Middle East Technical University, Ankara, Turkey, in 2000, and 2002 respectively. Towards her M.S. degree she worked with the Multimedia Research Group in Middle East Technical University. She is currently working toward her Ph.D. degree in Biomedical Engineering in the Imaging and Computer Vision Center, at Drexel University. Her main interests include sensor networks, stochastic modeling, detection and estimation theory, statistical pattern recognition, computer vision, and tissue characterization using ultrasound.

    View full text