Mobile and Pervasive Home > Mobile Service Discovery

Mobile Service Discovery

Overview

The research problem

Mobile devices have become portals to services beyond voice calls and text messaging. Because of this, wide network presence and technologically advanced mobile devices have a very central role to play towards the vision of pervasive and mobile computing. However, network providers are facing difficulties to increase data services usage up to satisfying levels. The aim of this project was to identify and address the potential reasons for this lack of usage.

In software engineering there are mainly two reasons affecting the success of a product: functional and non-functional requirements. These are respectively addressed with requirements engineering (what users want from a service/product) and usability design and evaluation (how users want to interact with it). However, in the mobile and pervasive environment there is at least one more significant factor which needs to be taken into consideration: service discovery. In this context, service discovery is addressed in its literal meaning, where humans discover services (and not as in the technical terminology where devices discover devices).

The future is foreseen with services which are context-dependent: available only in specific contexts. Such services already exist in the form of location-based or site-specific services. Similarly, services could be available depending on time, presence of other devices, social presence etc. This is expected to increase the problem of service awareness, as the set of available services will be changing dynamically.

Contribution

Researchers in context-aware computing have proposed many systems that attempt to infer users' intentions and suggest the relevant service(s) in every given context.

Instead of looking into how to improve service discovery through another context-aware system, the aim of this project was to investigate how service awareness can be facilitated in the first place. If the users know what services are available in every given context, they wouldn't need to attempt to discover them. This awareness can come through live notifications; for example, if you walk into an area where a service is available, your phone will make sure to notify you about this service (and of course annoy you immensely!). So, how can users be notified about services without annoying them?

This question led to research in the area of notification systems, ambient displays and multimodal interaction. The underlying assumption was that if satisfactory semantics in notifications are applied through all three available modalities (audio, visual, haptic), then they will either stand out or disappear in the background (pretty much as in the cocktail party effect). If the delivery mechanism is effective then notifications can be filtered out through a context aware system with a low threshold. If users are notified about irrelevant to the situation services, the meaningful notifications will make sure to absorb that extra annoyance. If none of the context aware systems is deemed satisfactory, it can be replaced with the traditional explicit input (e.g. user setting phone profile to "silent").