RSS and CSI combined fingerprint indoor positioning method based on deep learning

Abstract

The invention belongs to the technical field of wireless communication and indoor positioning. The invention discloses an RSS and CSI combined fingerprint indoor positioning method based on deep learning. The method comprises the following steps: collecting and processing RSS and CSI information in an offline stage, training a CSI processing model based on a deep auto-encoder, encoding CSI at a reference point, constructing a fingerprint library by combining RSS and CSI encoding, and dividing sub-fingerprint libraries according to an AP selection algorithm; training a positioning prediction model based on a deep neural network for each sub-fingerprint library; in the online positioning stage, according to the CSI processing model trained in the offline stage and based on the deep auto-encoder, carrying out CSI coding on the to-be-measured point, calculating joint RSS and CSI fingerprints, and carrying out prediction positioning based on a deep neural network model. According to the method, RSS and CSI information are combined, fingerprint information is enriched, and the calculation complexity and time consumption of a positioning prediction model are reduced; and the positioning precision is high, the expandability is strong, and the mobility is high.

Description

technical field [0001] The invention belongs to the technical field of wireless communication and indoor positioning, and in particular relates to a fingerprint indoor positioning method based on deep learning combined with RSS and CSI. Background technique [0002] At present, the closest existing technology: with the popularization of wireless networks and mobile devices, WLAN-based indoor positioning has received more and more attention. Its positioning accuracy can reach meter level, no need to deploy additional equipment, and the positioning cost is low. It is easy to promote in the application of daily life. WLAN-based indoor positioning technology can be divided into two types: ranging-based positioning algorithm (Ranging-based Localization) and location fingerprint-based indoor algorithm (Fingerprint-based Localization). Due to the complex and changeable indoor environment, the wireless signal is prone to reflection, refraction and scattering due to factors such as ...

AI Technical Summary

Problems solved by technology

The essence of DeepFi is to train a stacked restricted Boltzmann machine based on each reference point. When the positioning environment is larger and there are more reference points, the computational complexity and training time of the DeepFi algorithm will increase significantly.

On the other hand, the main idea of ​​the DeepFi localization method is to consider the abstraction of CSI at a single reference point, while at the same time, the connection between different reference points is ignored.

[0005] To sum up, the problems existing in the existing technology are: due to the coarse-grained and unstable characteristics of RSS, the positioning accuracy and stability based on RSS fingerprints are not high; and the redundant information and noise problems of CSI information have brought a negative impact on the positioning accuracy. challenge

However, CSI is a high-dimensional data, and there is redundant information between different subcarriers

If the CSI data is directly used for positioning, the calculation complexity of the positioning model will be high and it will be difficult to locate accurately

Images