CRII: CIF: Distributed Sensor Localization With Ordinal Data Constraints

点击翻译按钮获取中文摘要

This research involves the study of a new class of energy efficient optimization-based algorithms for localization in wireless sensor networks (WSNs) that can be used when conventional localization techniques such as GPS are not available (e.g. when WSN users are indoors). Such a system operates using only comparative distance measurements (as opposed to exact distance measurements) which precludes the need for strict clock synchronization among devices and enables heterogeneous cross-platform WSNs. Computation is implemented using the distributed capacity of the WSNs. The performance and energy efficiency of the algorithms are validated theoretically and by a combination of simulation and experimentation on an Android test bed. This research has diverse application areas such as health monitoring (location-aware patient care), security (first responders), and consumer electronics (location awareness in indoor malls).This research involves investigation into a new approach to localization in sensor networks where exact distance measurements are not required ? sensors need only comparative and ordinal distance measurements. This approach eliminates the need for strict clock synchronization between devices, or exact knowledge of operating conditions and environmental constants. It also eliminates the need for specialized hardware or external devices, making it possible to explore localization-based WSN applications on mobile phones and tablets in cross-platform environments. The research methodology involves the application of convex optimization to solve a Euclidean distance embedding problem. While such an approach has been used in areas such as image denoising and pattern recognition, it has never been proposed for localization in a WSN. Specific research topics include using optimization-based sensor localization, and parameter and function estimation via distributed computation. A testbed of Android devices is used to validate results.