Elsevier

Computer Communications

Volume 35, Issue 15, 1 September 2012, Pages 1921-1929
Computer Communications

An interactive cluster-based MDS localization scheme for multimedia information in wireless sensor networks

https://doi.org/10.1016/j.comcom.2012.05.002Get rights and content

Abstract

A wide range of applications used in wireless sensor networks requires location information of multimedia sensor nodes. In general, the topographical location information of data acquired by a sensor is applied for smart, interactive multimedia services. However, conventional techniques employ GPS or other location-tracking devices installed on sensor nodes and thus incur additional costs, making it impractical for wireless sensor networks. In contrast, some methods provide location information by node connectivity only. One of these methods, called multidimensional scaling – MAP (MDS-MAP), provides the most accurate positioning to date. However, MDS-MAP has a computational overhead of O(n3) in a network of n nodes and, in particular, results in significant localization accuracy error in environments with holes. Thus, this paper proposes a cluster-based MDS (CMDS) for range-free localization that overcomes the shortcomings of MDS and yields smaller accuracy error in all environments. Simulations demonstrate the proposed CMDS approach provides up to 23% improvement in localization accuracy compared to the newest version of conventional MDS-MAP, hierarchical MDS (HMDS) in a sensor network environment with holes.

Introduction

Localization techniques for wireless sensor networks use wireless communications among low-power, high-efficiency sensor nodes to indicate the location of each sensor node in an absolute or relative coordinate system. The future trend of media-aware content in ubiquitous systems is to require more location based services. Thus, it is very important for the sensor network and smart devices, such as smartphones or tablet PCs, to use the current location information. The localization of a sensor node is a priority requirement, as the current location information is a prerequisite to the provision of an environment in which a person could connect to the network at all times to obtain desired information [1], [2], [3]. Although a number of other proposed applications and techniques assume each sensor node has a GPS module or an additional location device capable of measuring absolute location, the use of GPS or an additional location device is fairly limited in an inexpensive sensor node with limited computational power. This leads to the proposal of a number of localization techniques for sensor nodes without additional locating devices.

Localization techniques for sensor nodes may be classified into: range-based techniques using GPS or other additional locating devices, and range-free techniques that do not use additional devices [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19]. Range-based techniques estimate location by the time of arrival (ToA) [20], which uses the travel time of data to measure the distance between nodes, the time difference of arrival (TDOA) [21], which uses the difference in transmission time of radio and ultrasonic signals to measure the distance, and the angle of arrival (AoA) [22], which uses the angle of received signals. However, these techniques require additional hardware on the sensor node to obtain distance or angle information for the sensor node and thus incur greater cost, making them impractical for real-world wireless sensor networks.

There are four well known range-free techniques. The first technique is the Centroid that receives location information from surrounding anchor nodes to perform the centroid calculation and then estimates location [23], [24], [25], [26]. The second technique, convex position estimation (CPE) [27], uses location information from neighboring anchor nodes and performs grid scanning. The third estimates localization by the approximate point in triangulation (APIT) [28], which uses triangulation including generated neighboring anchor nodes. The fourth (last) approach, multidimensional scaling-MAP (MDS-MAP) [29], estimates location using the connectivity information on all nodes. The anchor node refers to a sensor node with its own location information. Although the range-free techniques listed above perform localization using the connectivity among sensor nodes with no additional devices, message exchange is needed to collect connectivity information among sensor nodes, resulting in increased cost to calculate location. The range-free techniques have higher location accuracy error than range-based techniques. Conversely, MDS-MAP uses a matrix of distance information for all pairs of neighboring sensor nodes to obtain a relative coordinate and generates the most accurate location information among range-free techniques.

There is one significant drawback in MDS-MAP. The MDS-MAP yields higher location accuracy error if there are holes in the sensing field. A hole indicates a local area of a WSN which is incapable of transmission due to battery exhaustion or obstacles. The network hole causes higher localization errors. Yu et al. proposed a hierarchical MDS(HMDS) for overcoming the disadvantage of the MDS-MAP scheme. This scheme solves issues related to location accuracy error caused by holes using clusters. It divides the sensor topology into overlapped clusters. The cluster creation is important because the method for forming clusters and the size of cluster have influence on location accuracy error. If the size of cluster is big, the probability which is the estimated distance error between nodes is increased and the accuracy which is the result of MDS-MAP is decreased. However, HMDS overlooks this point and limits the number of clusters in simulation. These have resulted in lower location accuracy than our proposed scheme.

Thus, we propose a novel range-free localization scheme to overcome the drawback. It is called cluster-based MDS for localization (CMDS). The proposed CMDS form a number of k-hop clusters for localization. After forming clusters, each cluster gets its own coordinate system using MDS-MAP. Then CMDS gets one coordinate system, by merging coordinate system of all clusters. According to our simulation, the accuracy of the proposed CMDS and MDS-MAP is almost the same in WSNs without holes. However, holes usually exist in the real world of WSNs. Thus, we performed simulation in WSNs with holes; the proposed CMDS shows an improvement of up to 585% in localization error over MDS-MAP. In addition, our proposed CMDS gives superior accuracy performance with an improvement up to 23% compared to hierarchical MDS [30], one of the advanced MDS-MAP.

The remainder of this paper is organized as following: Section 2 discusses existing range-free techniques. Section 3 introduces the proposed cluster-based MDS-MAP (CMDS). Section 3 provides the simulation results of CMDS and analyzes the comparison to existing methods, HMDS and MDS. Finally, Section 5 concludes the paper.

Section snippets

Related work

This section explains basic localization schemes and related studies on MDS-MAP and HMDS. The Received Signal Strength Indicator (RSSI) is the most basic scheme among localization schemes. This scheme measures the distance between two sensor nodes using a characteristic in which the received signal strength is changed by the distance. A measurement of the distance using RSSI has the advantage of high practicality. However, the pattern of the signal strength is affected by the appearance of

Cluster-based MDS for range-free localization

This study proposes a localization technique in wireless sensor network environments where sensor nodes are randomly scattered in a large sensing field that requires monitoring. This paper has the following assumptions:

  • No sensor node has mobility.

  • Each sensor node has its unique ID, the same sensing range, and the same data transmission range.

  • At least one routing path exists between any pair of sensor nodes on the network. That is, all sensor nodes are connected.

  • Each sensor node is capable of

Performance evaluation

We use a Java simulation tool [37] to measure and compare localization error rates for MDS-MAP and HMDS to evaluate the performance of the proposed CMDS technique in this section. The simulation tool used in this experiment obtains a very similar result to the MDS-MAP schemes. Therefore, high reliability would be measured. The simulation environment is as follows: With a transmission radius of 1.5r, all sensor nodes are randomly placed in a 10 × 10r sensor topology, where r is a distance unit. We

Conclusion

In this paper, we propose CMDS that allows for accurate localization without additional positioning devices for multimedia services. Localization for multimedia services in a wireless sensor network using MDS-MAP results in a computational overhead of O(n3) and larger localization errors in an environment with holes. The proposed CMDS technique performs the MDS algorithm by locally generating clusters, resulting in a relatively lower computational overhead than MDS-MAP and improves localization

Acknowledgements

This research was supported in part by MKE and MEST, Korean government, under ITRC NIPA-2012-(H0301-12-3001), NGICDP(2011-0020517) and PRCP(2011-0018397) through NRF, respectively.

References (37)

  • I.F. Akyildiz et al.

    A survey on sensor networks

    IEEE Communications

    (2002)
  • S.H. Song et al.

    Enhancing location estimation and reducing computation using adaptive zone based K-NNSS algorithm

    KSII Transactions on Internet and Information Systems

    (2009)
  • D. Shin et al.

    Real-time tracking of human location and motion using cameras in a ubiquitous smart home

    KSII Transactions on Internet and Information Systems

    (2009)
  • J. Hightower et al.

    Location System for Ubiquitous Computing

    IEEE Computer Society

    (2001)
  • C. Wang et al.

    Sensor localization under limited measurement capabilities

    IEEE Network

    (2007)
  • J. Hightower et al.

    Location systems for ubiquitous computing

    Journal of Computers

    (2001)
  • D. Niculescu, B. Nath, Ad hoc positioning system (APS), in: Proceedings of IEEE Global Communications Conference...
  • J.P. Sheu, J.M. Li, C.S. Hsu, A distributed location estimating algorithm for wireless sensor networks, in: Proceedings...
  • L. Hu, D. Evans, Localization for mobile sensor networks, in: Proceedings of ACM International Conference on Mobile...
  • M. Rudafshani, S. Datta, Localization in wireless sensor networks, in: Proceedings of Information Processing in Sensor...
  • A. Baggio, K. Langendoen, Monte-Carlo localization for mobile wireless sensor networks, in: Second International...
  • J.P. Sheu et al.

    A distributed localization scheme for wireless sensor networks with improved grid-scan and vector-based refinement

    IEEE Transactions on Mobile Computing

    (2008)
  • Y. Wang et al.

    Range-free localization using expected hop progress in wireless sensor networks

    IEEE Transactions on Parallel and Distributed Systems

    (2009)
  • H. Mirehrahim, M. Dehghan, Monte Carlo localization of mobile sensor networks using the position information of...
  • S. Zhang et al.

    Accurate and energy-efficient range-free localization for mobile sensor networks

    IEEE Transactions on Mobile Computing

    (2010)
  • J. Sheu et al.

    Distributed localization scheme for mobile sensor networks

    IEEE Transactions on Mobile Computing

    (2010)
  • E. Kim et al.

    Mobile beacon-based 3D-localization with multidimensional scaling in large sensor networks

    IEEE Communications Letters

    (2010)
  • X. Wang et al.

    Component-based localization in sparse wireless networks

    IEEE/ACM Transactions on Networking

    (2011)
  • Cited by (44)

    View all citing articles on Scopus
    View full text