This is the 1st of several upcoming post about Context-Aware Mobile Computing, focusing on the area of Context Fusion especially.

Mobile Sensing is increasingly part of everyday life due to the rapid evolution of the mobile phone into a powerful sensing platform. Most consumer smartphones are now equipped with the necessary sensors to monitor a diverse range of human activities and commonly encountered contexts [1]. The sensing capabilities on mobile devices have the potential to enhance applications to provide useful services to users based on their situation. However, research has shown that most of the context-aware mobile applications developed are still lacking in utilising the sensing capabilities, and the key challenges of interpreting high-level and reliable context still remains unclear in mobile sensing research.

The advanced sensors embedded in the smartphones, has turned these devices into a powerful sensing platform. With the sensing capabilities, it may potentially provide valuable information from the users and identify their activities [1]. Researchers are currently looking at the area of mobile sensing in many different aspects, ranging from personal sensing to global sensing, from looking at using hard (physical) and soft (social network, web services) sensors, and investigating continuously sensing to maximise energy efficiency [2]. However, my research is only focusing on personal sensing (which is sensing people and their environment), due to the scope of my research project. Personal sensing applications are designed for single individuals, and are focused on collecting and analysing the data and the process of this is what this research is investigating.

As technologies shift toward mobility, Contextual Information becomes the most important piece of information and plays a major role in both mobile technologies and ubiquitous computing [3]. A system understands and using the context information around the mobile users, can provide a more natural way of interaction and can change their life significantly and by taking advantage of contextual information in a mobile environment (such as time and location) can provided great services to the users [4]. Dourish [7] categorised context into 2 different views; one is describing the setting of the where user is or whats around him, what he is doing. The other one is looking from a social science point of view, such as why is he doing this? Currently, I’m focusing on deriving context for mobile application, therefore is focusing on representational view of context. Moreover, research has shown most of the applications are only using low-level information (straight from sensor data)[6], and most of the applications believe the context they are using is completely accurate, but in reality context is often inaccurate and ambiguous [5]. Therefore, this research will investigate 2 aspects: Context fusion techniques: how sensor data can be derived into high-level context information for mobile applications to use, and Context Reliability: look into improving the accuracy of the derived context.

In order to test and development a framework for context fusion, I’m planning to undertake a case study on public transport. Transportation is a mobile activity, and mobile devices are well suited to sense and provide feedback about these activities. Mobile applications have the potentially to provide immediate feedback that may impact on how people travel on their daily transportation activities. The interaction of this case study along with mobile sensing and contextual information is my research focus. So the overall argument approach for context fusion has the potential.

References: