My master's thesis

For my master's thesis, I worked in a relatively new area in Computer Science called Pervasive Computing. In brief, Pervasive Computing deals with making use of the widely available computing power (available in PDAs, mobile phones, laptops, PCs, etc) along with the increasingly popular wireless networking technology to provide ubiquitous access to informaiton resources (i.e informaiton can be accessed from anywhere, at any time, and from a variety of devices.

But, however facinating the idea of Pervasive Computing is, it has many challenges and intricacies when it comes to implementation. One of the challenges is content adaptation. Since the various devices used to access information vary greatly in display size, processing capability, network bandwidth, storage capability, etc, it is difficult to present information for all of them in the same manner. Hence, some form of customization is required. There are various types of adaptation depending on the type of the information source to be adapted and the device/user/network it is going to be adapted to.

My thesis deals with the management of metadata for multimedia content. Metadata is basically data that describes another data. For instance, the information we find in library catalogue cards is an example of metadata. It is from the metadata that we can get appropriate information about the content without actually accessing the content. Content adaptation would be very much facilitated with the help of the content metadata. Hence, in my thesis, I have developed and architecture that facilitates the management, with respect to content adaptation, of metadata for multimedia content.

This thesis project was carried out in association with INSA, Lyon in France. There, a service-based adaptation framewrok is being developed by Girma Berhe who is finalyzing his Ph.D. My thesis is based on this framewrok and suppliments it by providing a management for the content metadata, which is required by the adaptation framewrok.

Abstract

One of the important aspects of a pervasive environment is the adaptation of content to suit a client's specific needs and choices such as the client's preferences, the characteristics of the client device, the characteristics of the network to which the client is currently connected, as well as other related factors. In order for the adaptation to be efficient while satisfying the client's requirements and maintaining the semantics and quality of the content, the adaptation system needs to have adequate information regarding the content to be adapted, the client's profile, the network profile and others. This thesis work addresses the issue of content metadata management in a pervasive environment in relation to content adaptation. For this purpose, a distributed architecture for the management of metadata of multimedia content is proposed. The proposed architecture consists of components for storage, retrieval, update, and removal of metadata in the system. It also provides interfaces to external components through which metadata can be accessed. In addition, restrictions on the adaptations that may be applied on the content are incorporated in the metadata. This enables the content creator to impose constraints on adaptations that may potentially cause the loss of critical information or a decrease in the quality of the adapted content.

Area: distributed systems, pervasive sytems
Keywords: metadata, pervasive systems, content adaptation, adaptation constraints, multimedia

Download

If you are interested, you can download the electronic copy of my thesis document below (in pdf format):

Multimedia Content's Metadata Management for Pervasive Environment

Areas of interest

I'm very interested in Distributed Systems and it's derivatives. It is a very interesting and challenging area of Computer Science which is still open for reasearch. Many specialized branches of Distributed Systems are being studied in various dimensions (for e.g Pervasive Computing and Grid Computing). For my master's thesis, I studied a very small issue in Pervasive Systems regarding content adaptation. By doing so, I've come to see that the area is very intersting and very open for researches. There are several issues that still need a lot of research such as security, for example. Hence, I would very much like to continue doing research in this area.

Another area of research that has come to capture my attention is the concept of integrating 'semantics' into things. It is currently difficult to automatically extract semantic features from content. As an example, we can easily extract the dimension (in pixels) of an image automatically using software tools, however we cannot automatically detect what the image contains (i.e is it a photo of a perosn, an X-ray of a patient's lung, a house, ...). I believe it is a tough thing to do but if appropriately studied and researched, it could be made possible.

I want to continue with my Ph. D. after a few years. I haven't yet decided on which of the above two areas I'm going to do my Ph. D. but I know for sure that it is going to be either one of the two. But I need to research each of them in further detail to make that decision.