Research‎ > ‎

Research Statement

Executive summary

Networked systems and the Internet have facilitated the generation of very large volumes of data across administrative boundaries, much of which is user-generated. In order to make sense of it, various data mining algorithms are employed, which in turn raises privacy and security concerns. The widespread adoption of the concept of utility computing paradigm (e.g. the cloud) in recent years has emphasised the need for privacy-preserving means of data publishing, storage and mining. Networked storage of large volumes of data has also posed storage reliability challenges. Closely tied with the exchange of data in distributed environments are models of trust, e.g. in the context of location, social networks.

My research interests span trust, privacy and security in distributed computing environments and cloud computing. I am particularly interested in innovative models of computational trust as well as trust models in highly emotive application scenarios. I am also interested in privacy and data security with respect to data mining and data publishing in large scale distributed systems, for example: how these fit into different XaaS models of cloud computing. In addition, I am interested in design perspectives of distributed systems, e.g. peer-to-peer networks.

My publication list is available on the Research Publications page.

Undergraduate research

During my undergraduate studies, I worked in a €2.8 million EU IST Framework Programme V funded digital cultural heritage project - ARCO - and its lightweight prototype (ARCOLite). ARCO was aimed at providing museums with useful technologies for digitising, managing and presenting virtual museum artefacts in virtual cultural environments. In order to reduce the total cost of ownership, a lightweight digital cultural heritage prototype (called ARCOLite) having similar functionalities was developed primarily as my undergraduate final year project. I co-authored a number of papers based on this work, including one demonstrating the use of ARCOLite in teaching, in a UNESCO journal on engineering education. Further to that, a portable version of ARCOLite was also developed which could run off a portable media. This was developed for educational purposes.

Doctoral (PhD) research  

My doctoral work looked into developing a reputation framework based on behavioural history of network clients, which can be used as means to control service levels to protect network services from abuses and unsolicited communication. Results showed that a policy-independent framework is useful at detecting (at the application layer) and thwarting many network service provisions where clients break their terms of service.

Alongside my doctoral research, I also worked in two EPSRC funded projects (£1.4m UTIFORO and £1m SHYNESS), both associated with computational trust models with direct relations to social implications of such trust models. As part of UTIFORO, I developed a trust-based shopping application in farmer's markets. Experiences from the trust model were used in developing models in SHYNESS.

In addition, I have worked on surveys of peer-to-peer network simulation tools. We surveyed a number of papers on peer-to-peer networking to find out the simulation tools they have used. We concluded that much of the peer-to-peer research uses custom simulators thus making their results irreproducible.

Post-doctoral research

In my recent work on privacy, I have collaborated with Rutgers University and British Telecom, producing a number of proposals of privacy preserving collaborative filtering schemes suitable for SaaS construction cloud environments such as the Google App Engine and the Amazon Elastic Beanstalk. Recently, I am also co-authoring a journal paper (under review) and an IETF draft on the state of simulations and simulators in peer-to-peer networks, which follows on from our earlier work in 2006-2007.

Another of my recent work on trust in highly emotive application scenarios reasons why traditional models for computational trust cannot be applied to a highly emotive scenario such as online dating.

Future research