Study On Positioning Algorithm Based On RSS Fingerprint In The Limiting Space

With the widespread use of Internet of Things, more and more wireless sensors are deployed. It is well known that sensors need not only perceive measured parameters but make clear their locations. That’s to say, the perceived information without locations is meaningless. Therefore, the accurate positioning of wireless sensors has been a key technical issue in the process of the development of Internet of Things technology. Among many methods of wireless position, the positioning method based on RSS fingerprint model has characteristics of the simple deployment, low hardware cost and wide range of application. And thus, these characteristics attract many researchers’ attention. However, the method has many problems such as maldistribution of positioning error, large workload to establish RSS fingerprint model and low positional accuracy. Therefore, in order to solve the above problems, the research involves the main parts:For the maldistribution problem of positioning error, distribution model of positioning error of localization algorithm based on RSS fingerprint model is study, which the rule of error distribution is got from. Division of different localization error area is obtained utilizing the rule of error distribution, the study lays the foundation for dividing LR area based on super-resolution principle, the rule is used simultaneously for further improving localization accuracy of positioning algorithm.For the problem of large workload to establish RSS fingerprint model and low accuracy, the RSS fingerprint rapid-generating algorithm based on Kriging is proposed. Firstly, the algorithm conducts the structural analysis to space field of signal strength in positioning area and choose theory variance function model on the premise of sufficient understanding of the nature of the field. Secondly, variance function is calculated by using observed value and relevant weight coefficients are gained on the condition of unbiased estimation and minimum variance estimation. Finally, RSS fingerprints on estimated points are calculated by Kriging estimator. Compared with the traditional method to gather RSS fingerprint point by point, a little RSS fingerprints gathered on observation points in the localization area are needed to estimate accurately RSS fingerprint on other estimated points. It can solve the problem of large workload to traditionally establish RSS fingerprint model and low accuracy.To solve the problem of low positional accuracy, a positioning algorithm based on SVM of FKC(fuzzy kernel clustering) is proposed, increasing technology of SVM is utilized under the condition of without raising computation complexity of positioning algorithm, fingerprint resolution and positional accuracy of localization algorithm are respectively improved, and the train process of SVM are quickened by fuzzy technology. For further improving positional accuracy of localization algorithm, positioning area is divided some LR scheme by positional error model on the basis of the above localization algorithm, a novel positioning algorithm based on super-resolution principle is proposed, and the improving maximum of positional accuracy of localization algorithm is 50%.The above research results have solved the existing problems of positioning method based on RSS fingerprint model, technical support for the accurate positioning of wireless sensor in the Internet of Things was not only provided, but also new idea for studying on accurate localization method of wireless.