text stringlengths 0 134 |
|---|
Challenges in Ubiquitous Data Management |
Improved hardware and networking are clearly central to the development of |
ubiquitous computing, but an equally important and difficult set of challenges revolve |
around Data Management [AK93]. In order for computing to fade into the |
background while supporting more and more activities, the data required to support |
those activities must be reliably and efficiently stored, queried, and delivered. |
Traditional approaches to data management such as caching, concurrency control, |
query processing, etc. need to be adapted to the requirements and restrictions of |
ubiquitous computing environments. These include resource limitations, varying and |
intermittent connectivity, mobile users, and dynamic collaborations. |
In this paper we first discuss the main characteristics of applications that |
ubiquitous computing aims to support and then focus on the requirements that such |
applications impose on data management technology. We then examine several |
different aspects of data management and how they are being adapted to these new |
requirements. |
Applications and Data Management Requirements |
While there is wide agreement on the great potential of ubiquitous computing, it is not |
yet clear what the killer applications (i.e., the uses that will result in widespread |
adoption) will be. Many researchers and product developers have created example |
scenarios to demonstrate the potential of the technology. Due to the integrated and |
universal nature of ubiquitous computing, these scenarios tend to include a large |
number of functions rather than any one single application. Thus, some in industry |
have begun to talk in terms of delivering a certain type of user experience rather |
than a particular application or suite of applications. These scenarios tend to involve |
users with several portable devices, moving between different environments (e.g., |
home, car, office, conference). The devices typically take an active (and often |
annoying) role in reminding the user of various appointments and tasks that are due, |
provide access to any and all information that may be relevant to these tasks, and |
facilitate communication among groups of individuals involved in the tasks. |
Categories of Functionality |
Rather than specify yet another such scenario, it is perhaps more useful to categorize |
the functionalities that such scenarios imply. This categorization can then be |
examined to determine the requirements that are imposed on data management. The |
functionalities can be classified into the following: |
1) Support for mobility the compactness of the devices combined with |
wireless communication means that the devices can be used in mobile |
situations. Thus, existing applications must be able to operate in varied |
and dynamic communication and computation environments, possibly moving from one network or service provider to another. Furthermore, |
new applications that are location-centric will also be developed. |
2) Context awareness if devices become truly ubiquitous, then they will |
be used constantly in a wide range of continually changing situations. |
For the devices to be truly helpful, they must be aware of the |
environment as well as the tasks that the user is performing or will be |
performing in the near future. Context aware applications range from |
intelligent notification systems that inform the user of (hopefully) |
important events or data, to smart spaces , that is, rooms or |
environments that adapt based on who is present and what they are |
doing. |
3) Support for collaboration another key theme of ubiquitous computing |
applications is the support of groups of people. This support consists of |
communications and conferencing as well as the storage, maintenance, |
delivery, and presentation of shared data. Collaborations may be |
performed in real-time, if all of the participants are available, or may be |
done asynchronously otherwise. In addition to supporting on-going |
collaboration, access to and analysis of traces of past activities is also |
required. |
Adaptivity and User Interaction |
These functionalities provide a host of challenges for data management techniques, |
but one requirement is present across all of them, namely, the need for adaptivity. |
Mobile users and devices, changing contexts, and dynamic groups all impose |
requirements for flexibility and responsiveness that are simply not addressed by most |
traditional data management techniques. Thus, adaptivity is a common theme of the |
techniques that we discuss in the remainder of the paper. |
It is also important to note that because ubiquitous computing is intended to |
augment human capabilities in the execution of various tasks, the nature of these |
applications is that the user is typically interacting in real-time with the computers. |
We are able to exploit this fact as part of the solution to adaptivity by, in some cases, |
depending on the users to make dynamic choices or to cope with some degree of |
ambiguity. A concrete example of such a design choice is the way that many |
groupware systems handle concurrent access and update to shared data. Rather than |
impose rules that restrict the types and degrees of interaction that users can have, as is |
done by concurrency control mechanisms in traditional database systems, a |
groupware data manager will typically impose less stringent rules. The relaxation of |
these rules limits the extent to which the system can autonomously handle conflicts. |
Thus, such systems typically handle whatever cases they can, and when they detect a |
conflict that cannot be handled automatically, they simply inform the user(s) that the |
conflict has occurred, and allow them to resolve it based on their knowledge of the |
situation. Thus, having users in the loop can be leveraged to provide more adaptive |
and flexible systems. |
Challenges in Ubiquitous Data Management |
Requirements Due to Mobility |
Other data management requirements are less universal across the three categories but |
yet must be addressed in order to support a comprehensive ubiquitous computing |
environment. For example, the issue of mobility raises a number of issues. First, the |
fact that the terminals (i.e. devices) are constantly moving, and often have limited |
storage capacity means that a ubiquitous computing system must be able to deliver |
End of preview. Expand
in Data Studio
README.md exists but content is empty.
- Downloads last month
- 4