CTIT University of Twente
CTIT Connecting Project

Mobile sensing: going beyond locations awareness

Title: Mobile sensing: Going beyond locations awareness

Description: Although the modern smartphones come with a variety of sensors, building-level location is by far the most utilized user information for enabling smartness in mobile applications [R1]. In this assignment, we will explore the possibility of utilizing both the hard sensors (accelerometer, barometer, camera, etc.) and soft sensors (calendar entries, application usage patterns, etc.) to recognize the current situation of the users and their activities. The possible applications include,


A Pedestrian Dead Reckoning (PDR) app that utilizes inertial sensors such as accelerometer and gyroscope for indoor navigation [R2].


Inertial sensors based sleep analysis app that can measure the quality of sleep [R3].


An indoor localization app which leverages the visible WiFi access points and the signal strengths to localize the users at floor (or) room level.


A lifelogging application for tracking the social activities of the users based on their location and noise level.

Students can work on various scientific challenges depending on their interest and expertise. Some typical challenges are,


Improving the accuracy of the recognized user situations without draining the battery of the mobile devices. Students will investigate the use of various user related information, signal processing techniques and machine learning algorithms for reducing the number of false positives.


As most of the above described applications rely on machine learning algorithms, they require the active participation of the users for training and to provide corrective feedbacks to these algorithms. Students with interest in HCI (Human Computer Interaction) can explore the suitable methods to provide feedbacks to the users based on the probabilistic outputs of these algorithms and techniques to effectively capture the user preferences.


With the possibility for tapping into mobile sensor data, there is an increasing risk of unauthorised device identification and tracking leading to privacy and security issues. Hence, a possible research track is to explore the feasibility of uniquely identifying and tracking the mobile devices with readily available sensor data such as accelerometer values or battery power consumption [R4].

The students can choose an application and work on one or more challenges depending on their interest and study requirements. In all cases, a simple working prototype will be provided at the beginning. Also, the students are encouraged to propose their own smart mobile applications that can leverage sensors available on the mobile devices.


(1) Good programming skills in Java. Familiarity with Android programming is an advantage. (2) Interests in mobile application development and machine learning techniques.

Level: Bachelor/ Master

Contact: Arun kishore Ramakrishnan, r.arunkishore@gmail.com


[R1] Christos Emmanouilidis et al., “Mobile guides: Taxonomy of architectures, context awareness, technologies and applications”, Journal of Network and Computer Applications 2013, url: http://www.sciencedirect.com/science/article/pii/S1084804512001002

[R2] Yuya Murata et al., "Pedestrian Dead Reckoning Based on Human Activity Sensing Knowledge", UBICOMP '14 ADJUNCT, url: http://ubicomp.org/ubicomp2014/proceedings/ubicomp_adjunct/workshops/HASCA/p797-murata.pdf

[R3] Ondrej Krejcar et al., "Use of Mobile Phones as Intelligent Sensors for Sound Input Analysis and Sleep State Detection", Sensors 2011,

url: http://www.mdpi.com/1424-8220/11/6/6037

[R4] Yan Michalevsky et al., "PowerSpy: Location Tracking using Mobile Device Power Analysis", Usenix Security 2015, Url: http://arxiv.org/abs/1502.03182