Early detection of a fall using wi-fi and deep learning | |
| Author | Warayut Surasakhon |
| Call Number | AIT Thesis no.TC-20-02 |
| Subject(s) | Orthogonal frequency division multiplexing MIMO systems Machine learning Falls (Accidents) in old age |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Telecommunications |
| Publisher | Asian Institute of Technology |
| Abstract | Falling is the important cause of injury and death in elder. In 2015, there are about 226 million cases of dramatic falls and 527,000 people died. While the problem is more serious, the number of elder is continuously increasing. To relieve this issue, timely treatment is important, so we present fall detection utilizing Wi-Fi channel state information (CSI). The CSI provides information about the environment which the signals travel through. Therefore, different human movements impact CSI differently. In this thesis, first, we process the raw CSI to remove the part which is not caused by human movement. Next, we propose data preparation and feature selection in order to eliminate bias resulted from poor data quality. Moreover, we perform experiments to realize a criterion of device installation that provides the excellent results. To interpret human activities, we employ several classification algorithms to observe which algorithm is the best for fall detection. We offer not only basic classification models consisting of support vector machines (SVM), K-nearest neighbor (KNN), decision tree, random forest, and naive Bayes, but also deep learning models consisting of fully-connected network and long short term memory network (LSTM). The result of the experiments show that fully-connected network and random forest are the top 2 models which can correctly detect falling with accuracy of 97.7 and 95.1 percent in single room, and 92 and 92.4 percent in two rooms with wall, respectively. |
| Year | 2020 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Telecommunications (TC) |
| Chairperson(s) | Attaphongse Taparugssanagorn; |
| Examination Committee(s) | Teerapat Sanguankotchakorn;Poompat Saengudomlert; |
| Scholarship Donor(s) | His Majesty the King’s Scholarships (Thailand); |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2020 |