title: Behavioral and Functional Partitioning of Digital Systems
reg no: ETF5601
project type: Estonian Science Foundation research grant
subject: 2.9. System Engineering and Computer Technology
status: completed
institution: TTU Faculty of Information Technology
head of project: Peeter Ellervee
duration: 01.01.2003 - 31.12.2005
description: The main goal the project is to develop new methods, algorithms and software for digital systems partitioning based on their behavioral and/or functional descriptions.
The main tasks are as follows:
1) Development of partitioning methods for algorithms at behavioral level with a goal to reduce data-transfer between components.
2) Partitioning of circuit's components (modules) starting from an algorithm at register transfer level, i.e. partitioning of single modules depending on their functionality.
3) Integration of both partitioning methods into a unified methodology, and development of related estimation methods.
4) Development of heuristic and iterative partitioning algorithms and implementing them in a prototype design environment xTractor.
The following results are expected:
1) The reduce of data-transfer between components should be followed also by the reduce of total number of interconnections and of power consumption. Focusing onto data-transfer helps to avoid dependencies on details that will be available at lower abstraction levels.
2) Partitioning at register transfer level will allow to take into account physical level parameters when generating corresponding components. The result will be significantly more optimal when comparing with partitioning at algorithmic level but optimization is also more complex computationally.
3) Combining both partitioning methods would allow, first, to explore architectural solutions and, second, to achieve more accurate final results.
The heuristic partitioning methods will be based at fast heuristic methods that were developed during earlier research for weighted graph coloring. Iterative methods will be based on various neural net and genetic optimization algorithms.

project group
no name institution position  
1.Peeter EllerveeTTU Faculty of Information TechnologyAssoc. Professor 
2.Margus KruusTallinn Technical UniversityAssoc.Professor 
3.Aimar LiiverEMTengineer 
4.Kalle TammemäeTallinn Technical UniversityAssoc.Professor