Research Interests: Ken MacGregor

Prof. MacGregor’s area of research covers distributed computing in all its aspects, that is, the distribution of applications between different computer systems, irrespective of the network type. With the growth of network technology especially the Internet andwireless communications, the use and development of distributed applications in rapidly increasing, however tools for the development and analysis of such applications are still not properly understood. The techniques used to develop commercial client/server systems across corporate networks are not necessarily suitable to implement distributed applications over inherently unreliable networks. In South Africa with its wide geographic distribution of population and shortage of skills this area of computing has many particular sociological advantages and challenges. The research covers three specific areas in which Prof MacGregor is looking for PhD and MSc students. These are:

Wireless Middleware Middleware is the name given to the software that enables the communication between distributed applications. In a wireless world where connections can be intermittent and unreliable, a greater emphasis is placed on the functioning of the middleware. Different requirements exist for synchronisation of diaries in wireless devices as distinct from nonidempotent transactions. This research considers the requirements of wireless middleware and its production.

Open Source Operating Systems The purpose of this research is to adapt the available Open Source Software and make it more user friendly and generally usable. The research is currently looking at a number of areas:

• Operating Systems the LINUX family from full LINUX on a server to mLINUX on a PDA or cell phone.
• Application development tools for ease of usage
• Desktop software

Peer-to-Peer File Systems The increasing availability of bandwidth presents new opportunities for distributed computing. Instead of the traditional message passing paradigm, which puts an additional overhead on the application, the more natural peer-to-peer paradigm could be adopted. In this method of distribution the application can discover what file resources are available on other computers and access them as if they were on their own systems. In this type of environmentmany user applications can cooperate accessing this common shared file system. The project is implementing a peer-to-peer wireless file system, consider its limitations, investigating the algorithms for ensuring the concurrency and integrity of the data, and researching the effects of bandwidth on the implementation. The software is being developed on a standard Open Source operating.