Introduction
Technological advancements have facilitated the design and building of ad hoc sensor networks that have uses in a range of applications. Due to their low power processors, inexpensive nodes, modest memory and wireless capability, ad hoc sensor devices have been used in a number of novel applications such as habitat monitoring, target tracking, smart building failure detection and even in military. These applications necessitate require accurate orientation of the nodes with respect to the global coordinate systems in order to derive meaningful data. This paper attempts to explore some of the localization techniques that have been adopted including Monte Carlo localization, RF localization technique, LSVM and other proposed localization logarithms.
In spite their apparent simplicity and operational requirements, they present novel tradeoffs in design and use. On one side, their low cost of production facilitate massive scale and highly parallel computations. On the other, each node is limited by power and reliability and communication with the next node. Many sensor network applications require location awareness in order to function effectively. However, it is always expensive to include GPS receivers in the network sensors, thus further limiting their use.
Most localization techniques developed have attempted to gather information into a central location upon which they are analyzed using mathematical algorithms. Semi-definite programming and multidimensional scaling techniques are excellent but not applicable on a large scale. Likewise, relaxation-based techniques and coordinate-system stitching lack accuracy. According to , accuracy in LSVM can be achieved using beacon-based techniques. In this method, known nodes known as beacons are extrapolated to determine unknown node locations. Most current techniques assume that the distance between two neighbour nodes can be determined using ranging processes. For example, pair wise distances can be determined via RSSI, AoA or TDoA. The problem with ranging processes is that measurements are subjected to noise and its complexity/cost increases with accuracy needs.
In sequence-based localization, where the localization space can be divided into distinct regions represented by O(nn) sequences, RF signals are used since radios are used for communication in almost all the devices in a wireless network. Ideally, the computed distance of the reference nodes should be identical to the distance order determined by Euclidean means. However, this is not the case in reality as RF signals are subjected to multipath fading and noise. The non-ideal effects alter the location sequence measured by the unknown nodes such that instead of O(nn) combinations of distance rank sequences, only O(n4) are possible.
Problem identification
Localization algorithm designs have some apparent issues that include resource constraints, node density, non-c0nvex topologies and environmental obstacles. Localization algorithms find difficulty in locating nodes near the edges of a sensor field. This is known as non-convex topologies. In terms of resources, nodes have weak processors, restricting large computations. Environmental obstacles cannot be ruled out since large rocks and uneven terrains occlude line of sight, preventing TDoA ranging, radio interference and errors in RSSI thus resulting in incorrect hop count.
A particular localization that has been proposed to improve precision and accuracy of localization is Monte Carlo Localization In mobile applications. In this technique, problems such as requirement for special software and particular network topologies are surpassed by designing a network of static nodes and moving seeds and vice versa. In this method, it is found that mobility can improve accuracy and decrease the cost of localization.
Conclusion
In conclusion, for wireless sensor networks to accurately determine locations they require nodes. Different localization models have been proposed and while some are not perfect, others such as Monte Carlo localization seem to eradicate most of the problems associated with other models. This includes cost, accuracy and precision and computational power.
Works Cited
Duc A. Tran, and Thinh Nguyen. Localization In Wireless Sensor Networks based on Support Vector Machines. IEEE, n.d.
Jonathan Bachrach, Christopher Taylor. Localization in Sensor Networks. MIT, 2009.
Krishnamachari, Kiran Yedavalli and Bhaskar. Sequence-Based Localization. IEEE, 2008.
L. Doherty, L. E. Ghaoui, and K. S. J. Pister,. “Convex position estimation in wireless sensor networks,” in IEEE Infocom. April 2001.
Lingxuan Hu, David Evans. Localization for Mobile Sensor Networks. In Tenth Annual International Conference on Mobile Computing and Networking, 2004.
Mazzini., P. Bergamo and G. Localization in Sensor Networks with Fading and Mobility. IEEE PRMC, 2002.
S. Ray, D. Starobinski, A. Trachtenberg, and R. Ungrangsi,. “Robust Location Detection with Sensor Networks,” IEEE J.Selected Areas in Comm.,special issue on fundamental performance limits of wireless sensor networks. 2004.
T. He, B.B.C. Huang, J. Stankovic, and T. Abdelzaher,. “Range-Free Localization Schemes for Large Scale Sensor Networks,”. ACM, 2003.
