WLAN Indoor Positioning Algorithm Based On Locally Embedding And Neural Network

The problem of dimensionality reduction arises in many fields of information processing, including machine learning, data compression, scientific visualization, pattern recognition, and neural computation. Nowadays, due to the positioning algorithm of indoor WLAN stores Radio map data with too many dimensions in the offline positioning. In the case of doing not need the entire map and needing real-time update, the large amount of data becomes the problem of real-time downloading.Firstly, this paper analyzes locally linear embedding(LLE), an unsupervised learning algorithm that computes low dimensional, neighborhood preserving embeddings of high dimensional data. Notably, the optimizations in LLE—though capable of generating highly nonlinear embeddings—are simple to implement, and they do not involve local minima. In this paper, we describe the implementation of the algorithm in detail and discuss several extensions that enhance its performance. We present results of the algorithm applied to data sampled from known manifolds, as well as to collections of images of faces, lips, and handwritten digits. Finally, this paper give the simulation of this algorithm parameters.Secondly, Aiming at the problem of large amount of real-time data transmission, the paper proposes a method that we use the BP network to compress at the sending end, and then save the weight. The method reconstructs data in the terminal, which greatly reduces the amount of data in real-time transmission to ensure the fast and effective of real-time positioning. The used simulation environment is in the 27 AP in the floor for data acquisition and real-time simulation application. The applied algorithm is a universal KNN algorithm. The core data compression method is the improved BP neural network, give BP network training and simulation.Finally, focusing on the data in the input end of BP network, the paper makes the real-time Radio map be normalized and presets the simulation parameters. The simulation results show that in the case of choosing appropriate parameters to simulate, compared to the RAR compression it can compress the whole Radiomap ion to 1/2 or even 1/3 of data size, it can maintain high stability in a certain degree of compression, and it can guarantee the reconstructed and compressed data has lossy compression within the allowable range of positioning accuracy, combined with semi-supervised algorithm, give the best program.