SenseMe: A System for Continuous, On-Device, and Multi-dimensional Context and Activity Recognition

Abstract

In order to make context-aware systems more effective and provide timely, personalized and relevant information to a user, the context or situation of the user must be clearly defined along several dimensions. To this end, the system needs to simultaneously recognize multiple dimensions of the user's situation such as location, physical activity etc. in an automated and unobtrusive manner. In this paper, we present SenseMe-a system that leverages a user's smartphone and its multiple sensors in order to perform continuous, on-device, and multi-dimensional context and activity recognition. It recognizes five dimensions of a user's situation in a robust, automated, scalable, power efficient and non-invasive manner to paint a context-rich picture of the user. We evaluate SenseMe against several metrics with the aid of 2 two-week long live deployments involving 15 participants. We demonstrate improved or comparable accuracy with respect to existing systems without requiring any user calibration or input.

A Location-Based Interactive Model of Internet of Things and Cloud (IoT-Cloud) for Mobile Cloud Computing Applications

“…The accuracy of the current localization techniques is good enough for our model as well as similar existing location-based applications like mobility pattern monitoring of moving objects in large cities [22], continuous multi-dimensional context and activity recognition [23], criminal tracking, autonomous car navigation and obstacle avoidance [24,25]. Real-world data from the Federal Aviation Administration show that their GPS attains better than 2.168 m accuracy with a 95% confidence level [26].…”

Modeling Users’ Behavior from Large Scale Smartphone Data Collection

“…computed over all magnitudes of sensor values captured over a 1-second window. Hence, fine-grained activity recognition as performed in [5,20,21] is not possible with this dataset.…”

“…• Bluetooth count: As suggested by our previous work in [5], bluetooth count can be utilized to determine the number of people around. We compute an average bluetooth count for each unique place in the user's location history.…”

“…While previous work in activity recognition (including SenseMe [5], CenceMe [21] and Jigsaw [20]) has focused on fine-grained activity recognition such as Walking, Running, Driving etc. , it is not possible to do that with the DeviceAnalyzer dataset because of the limited accelerometer data available and its sampling rate (see Section 4.1).…”

“…However, a large volume of research in mobile and ubiquitous systems has been devoted to using this collected information for inferring users' current high level context such as their environmental context (whether they are indoors or outdoors), the activities they are engaged in (walking, driving etc.) and how many people are around them [5,20,21]. On the other hand, mining of users' diverse longitudinal behavioral patterns from rich smartphone data, which can enable exciting new context-aware applications and services, has not received much attention.…”