University of Twente Student Theses


Group Activity Recognition Using Channel State Information

Visserman, H.A. (2019) Group Activity Recognition Using Channel State Information.

[img] PDF
Abstract:Activity recognition using WiFi signals can offer a good way to identify the activities of a group of performers on stage. Using device free activity recognition, the performance does not need to be recorded nor do the participants need to wear sensors to recognize the performed activities. Research has been performed regarding the recognition of single person activities using the channel state information (CSI) of WiFi signals. In this research it is shown to what extend one can identify what activity the biggest part of a group of people is performing using CSI. To answer this question, the performance of two different machine learning algorithms, decision tree and support vector machine, are compared under different circumstances. The varying conditions are the percentage of the group performing the main activity, and the amount of nodes used to receive the WiFi signal. The results show that it is possible to track the activity of a group of participants using WiFi signals. The highest accuracy and F1 score of 98% and 0.94, respectively, were achieved using three nodes and the decision tree classiffer, when 75% of the group was performing the main activity.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science BSc (56964)
Awards:Best paper award in the track Ubiquitous Computing and IoT awarded at the 30th Bianual Twente Student Conference on IT
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page