An interactive cluster-based MDS localization scheme for multimedia information in wireless sensor networks Minhan Shon a , Minho Jo b,⇑ , Hyunseung Choo a,⇑ a School of Information and Communication Engineering, Sungkyunkwan University, Suwon 440-746, South Korea b School of Information and Communications, Korea University, Seoul 136-701, South Korea article info Article history: Available online 15 May 2012 Keywords: Clustering MDS Localization Range-free WSN abstract A wide range of applications used in wireless sensor networks requires location information of multime- dia 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), pro- vides the most accurate positioning to date. However, MDS-MAP has a computational overhead of O(n 3 ) in a network of n nodes and, in particular, results in significant localization accuracy error in environ- ments 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. Ó 2012 Published by Elsevier B.V. 1. Introduction Localization techniques for wireless sensor networks use wire- less 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 cur- rent location information. The localization of a sensor node is a pri- ority 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 infor- mation [1–3]. 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 de- vices, and range-free techniques that do not use additional devices [4–19]. 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 tech- nique is the Centroid that receives location information from sur- rounding anchor nodes to perform the centroid calculation and then estimates location [23–26]. The second technique, convex po- sition estimation (CPE) [27], uses location information from neigh- boring anchor nodes and performs grid scanning. The third estimates localization by the approximate point in triangulation (APIT) [28], which uses triangulation including generated neigh- boring anchor nodes. The fourth (last) approach, multidimensional scaling-MAP (MDS-MAP) [29], estimates location using the con- nectivity information on all nodes. The anchor node refers to a sen- sor 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, 0140-3664/$ - see front matter Ó 2012 Published by Elsevier B.V. http://dx.doi.org/10.1016/j.comcom.2012.05.002 ⇑ Corresponding authors. Tel.: +82 2 3290 4764, +82 31 290 7145. E-mail addresses: minari95@skku.edu (M. Shon), minhojo@korea.ac.kr (M. Jo), choo@skku.edu (H. Choo). Computer Communications 35 (2012) 1921–1929 Contents lists available at SciVerse ScienceDirect Computer Communications journal homepage: www.elsevier.com/locate/comcom