[time]
2022-09-02T02:37:15+02:00

[track]
3

[team_name]
MSCAS

[team_institution]
HUST

[logolink]


[team_members]


[reference_person]
mengweifei

[reference_email]
vicmengcs@163.com

[description_short]
Our initial idea is to combine WiFi RSSI positioning with step direction estimation positioning. In positioning, we will first use WiFi RSSI wireless fingerprint positioning to build a rough floor estimation of the building, and then use accelerometers and azimuth sensors to estimate the pace, so as to approach the real physical location.
Our method can be divided into three steps: The first step is data preprocessing. We need to convert the original data into data suitable for our method processing, so as to facilitate the subsequent deep neural network construction and training. The second step is to reduce the dimension of data. Since the input data is polymorphic and multidimensional, we have built an SAE(Stacked AutoEncoder) network to reduce the dimension of data. The last step is multi classification of data. After the dimension of the input data is successfully reduced, the specific position needs to be determined by using the input data. The input data is the WiFi RS

[description_long_link]
https://gitee.com/vicmengcs/ipin/blob/master/IPIN2022 competition .pdf

[publish_check_]
true

[results_check_]
true

[data_check_]
true

[pdf_check_]
true

[video_check_]
true