Purpose
Recently, many researches have been devoted to studying the possibility of using wireless signals of the Wi-Fi networks in human-gesture recognition. They focus on classifying gestures despite who is performing them, and only a few of the previous work make use of the wireless channel state information in identifying humans. This paper aims to recognize different humans and their multiple gestures in an indoor environment.
Design/methodology/approach
The authors designed a gesture recognition system that consists of channel state information data collection, preprocessing, features extraction and classification to guess the human and the gesture in the vicinity of a Wi-Fi-enabled device with modified Wi-Fi-device driver to collect the channel state information, and process it in real time.
Findings
The proposed system proved to work well for different humans and different gestures with an accuracy that ranges from 87 per cent for multiple humans and multiple gestures to 98 per cent for individual humans’ gesture recognition.
Originality/value
This paper used new preprocessing and filtering techniques, proposed new features to be extracted from the data and new classification method that have not been used in this field before.
Towards execution time estimation in abstract machine-based languages
