An approach for discovering inherent parallelism in process model structure and for combination of the processes
Taras Grytsenko graduated from Kharkiv National University of Radio Electronics, Ukraine and successfully passed his MPhil Viva at WIT with a thesis on “An approach for discovering inherent parallelism in process model structure and for combination of the processes”. The external examiner was Prof Vladimir Lovitskii from Natural Response TM, Bolton Institute, UK and the internal was Dr Viktor Popov.
The objective of the research was to develop an approach for discovering parallelism in a model structure and for the combination of any selected processes in a model.
This research has focused on analysing information processes represented in the form of mathematical circuits. Taras’ research investigated formal tools for representing process models: flow graphs for processes and basic process algebra operations.
The problem of the distributed simulation of complex information systems was investigated and stages of the original models reorganisation for distributed modelling were proposed. In addition an algorithm for discovering inherent parallelism in the structure of process models was developed.
Subsequently, the problem of adaptation of the original model to certain hardware platforms was investigated. As a result an algorithm for merging several processes into one uniform process was proposed. This merging operation was named the convolution operation.
Finally, a software tool based on proposed approaches was developed. To develop such a program Taras used the C++ language and the STL library. To estimate possible profit of using developed approaches and software, a special computer model was developed. The results obtained during modelling proved that the developed approach for discovering parallelism in a model structure, together with the operation for the combination of the processes gave impressive results in practice.
The results from this research may have applications in distributed discrete event simulation systems based on process models. With the help of the proposed approaches a source process model can be adapted for the distributed simulation on a base of a multiprocessor computer.
Taras presented an excellent MPhil thesis as a result of his research and was congratulated by both examiners.
The programme was supported through the Chevening Scholarship Scheme.