A multi-network assisted positioning method for distributed large-scale multi-antenna systems

Abstract

The present invention proposes a multi-network assisted positioning method for distributed large-scale multi-antenna system, a diffraction model for feature acquisition is introduced into the RSS-based DBN positioning, and the new RSS feature effectively considers other UEs The impact on the target UE improves the adaptability of the trained DLBP network in multi-user scenarios; LSTMN is used to analyze the motion trajectory of the UE with only a small amount of trajectory information, and by using the DBN estimation results and a small number of location samples, The complete trajectory of the UE can be effectively obtained. Compared with the traditional method, the cost of historical trajectory information collection and measurement is greatly reduced; the method combining the DBN and LSTMN estimation results through BPNN can achieve higher positioning accuracy.

Description

technical field [0001] The invention relates to the technical fields of wireless communication, positioning and machine learning, in particular to a multi-network assisted positioning method for a distributed large-scale multi-antenna system. Background technique [0002] With the development of the Fifth Generation (5G) network, the location information of terminal devices can be used to provide regional advertisements, content caching, and personnel tracking services under emergency calls, which makes wireless user location technology become an important part of the academic and industry circles. One of the research directions [1] Zhang Zixuan, Huang Jinan, Cai Zihua. Analysis of the development trend of 5G communication and positioning integrated network [J]. Guangdong Communication Technology, 2019,39(02):45-49. At present, a common outdoor communication system mainly adopts a satellite-based Global Positioning System (Global Positioning System, GPS) to acquire the locat...

AI Technical Summary

Problems solved by technology

Nevertheless, the existing solutions [17][18][19][20] require a large amount of historical trajectory information to estimate the next location of the target UE, so it requires high data measurement costs, which is not conducive to practical applications

[0007] In the real environment, with the continuous improvement of users' requirements for positioning accuracy, the training database is also getting larger and larger, which brings additional preliminary work to data collection, so it is difficult to adapt to the rapidly changing scene and high sampling cost. Application scenarios [21] T.Liu, Y.Yang, G.Huang, Y.K.Yeo, and Z.Lin, Driver distraction detection using semi-supervised machine learning [J], IEEE Transactions on Intelligent Transportation Systems, vol.17, no.4 , pp.1108–1120, Apr.2016

For example, in an indoor environment, the movement of indoor facilities such as furniture and objects and the movement of people will cause multipath and shadow effects in signal propagation, resulting in strong time-varying characteristics of RSS, which makes it impossible to analyze the same location after the training data set is collected. The sampling data can be reused for a long time

In the absence of regular update and maintenance, this will cause the problem that the positioning results gradually deviate from the accurate value [22] Li Yanjun, Xu Kaifeng, Shao Jianji. Research on the method of updating Wi-Fi indoor positioning fingerprint database by crowdsourcing [J]. Journal of Sensitive Technology, 2014(12):108-114

Traditional DL-based positioning algorithms do not fully consider the impact of mobile personnel on the RSS of target users in multi-user scenarios, fail to make full use of trajectory information, and require high sampling data collection costs, so they cannot fully meet the positioning needs of real environments. DL-based The positioning algorithm still needs further research

Images