Research Group:
Cellular Algorithms and Architectures
Head of the Group: O.Bandman, Doctor of Sciences.
A research group for fine-grained parallelism and cellular
architecture study consists of four researchers and works as a part
of the Parallel Algorithms and Architecture Laboratory. The main research
goal is to develop fundamental concepts, formal methods and
computer tools for designing algorithmic oriented architecture
for data array processing.
A formal model called Parallel Substitution Algorithm (PSA) has
been proposed, which conceptually ascends to fon-Neumann cellular automaton,
but has more powerful expressive capability.
The main peculiarities of the model are the following:
- in that computation space it is introduced in explicit form so that
the structure of intercell exchanges is expressed in the form of functions
determined in this space,
- the concept of context allows to express the controlling function directly
in the representation of the algorithm. Based on PSA concepts a theory
has been developed which comprise the correctness conditions, equivalent
transformations and a number of methods for algorithm and architecture
synthesis. A computer simulation system was elaborated, which allows to
construct cellular algorithms and observe computation processes in dynamics.
The following researches and projects are under way:
- cellular technology of computation in optical systems,
- neural networks with cellular architecture (Cellular Neural Networks),
- cellular arrays for fast arithmetic,
- universal cellular programmable arrays with multilayer architecture.
Main project: "Formal model of distributed computing /parallel
substitution algorithms/" (Supported by the Russian Foundation of
Fundamental Research under grant N93-01-01000).
Publications:
- O.L.Bandman, S.G.Pudov, A.V.Selikhov, Discrete cellular-neural networks
for image processing, (LaTex format) mobil.zip
- O.L.Bandman, S.G.Pudov, Stability of Stored Patterns in Cellular-Neural
Associative memory, (LaTex format) stab.zip
- S.Pudov, Influence of Self Connection Weight on Cellular-Neural
Network Stability, (LaTex format) self.zip
- A.Selikhov, Cellular Algorithm of Contour Extraction from a 2D Image,
(LaTex format) isoline.zip
Last update: October 18, 1997