RSSI Wireless Indoor Location Technology Based On Machine Learning

With the continuous development of the Internet of Things and artificial intelligence,wireless indoor positioning has gradually become a current research hotspot.Wireless sensor networks have the advantages of low cost and easy deployment,and are widely used in indoor positioning.The RSSI-based wireless positioning method does not need to add additional hardware and the data is easy to collect,which has also attracted people's attention.However,compared with other methods,the ranging error of RSSI is large.Therefore,how to achieve accurate position estimation is one of the challenges currently facing indoor positioning.Machine learning methods can make use of the model's learning capabilities to make predictions about the coordinates of unknown nodes,and can better adapt to various network environments.At the same time,due to different application scenarios,two-dimensional positioning cannot fully meet the actual needs.Therefore,the three-dimensional positioning method combined with machine learning has extremely important research value and practical significance.This paper introduces the basic ideas and deficiencies of machine learning methods in indoor positioning applications.At the same time,it studies several key factors that affect the positioning accuracy of wireless sensor networks and proposes an RSSI three-dimensional collaborative positioning algorithm based on the Light GBM model.Firstly,for the problems of large edge region positioning errors caused by factors such as beacon node distribution and sensor network morphology,and the incompleteness of feature vectors caused by the sparseness of RSSI data,which reduces the prediction accuracy of models such as SVM,this paper proposes Light GBM model's3D rough positioning algorithm.This method can better adapt to different network shapes and effectively improve the problem of uneven positioning error distribution.Secondly,Aiming at the problem that the collaborative node's own error affects the accuracy of correction,a collaborative node selection method based on accuracy strategy is proposed.This method includes a threshold screening strategy,a subset judgment strategy,and an anchor node replacement strategy.It can effectively screen out nodes with higher accuracy,thereby reducing the impact of errors in the coordination processon the correction accuracy.Finally,in view of the lack of accuracy of a single machine learning localization method and the inadequate use of information between unknown nodes for cooperation,a fusion and cooperation correction algorithm based on QSC and DVMFL is proposed.The method includes an improved centroid algorithm and a DV-Hop algorithm to further improve the accuracy of 3D positioning.Experiments show that the method in this paper is highly adaptable to various factors affecting the positioning accuracy of wireless sensor networks,and effectively improves the accuracy of 3D positioning.