Execution of distributed reactive systems

Abstract

This paper addresses the problem of distributed reactive systems execution. We first show that a natural parallel description of such systems can be achieved with synchronous languages. Then, we explain how a centralized synchronous program can be executed in its environment, which is intrinsically asynchronous. For this purpose, we define a synchronous/asynchronous interface, which links the program logical time with the environment physical time. Finally, we motivate the need for distribution and show how a desired distribution can be easily achieved, thanks to the object code distribution algorithm implemented in the ocrep tool. We then propose and discuss three solutions that allow distributed synchronous programs to be executed on an asynchronous network of processors.

BibTeX entry

@InProceedings{CG95,
  author = 	 {P. Caspi and A. Girault},
  title = 	 {Execution of distributed reactive systems},
  booktitle = 	 {First International Conference on Parallel Processing, 
		  EURO-PAR'95},
  editor = 	 {S. Haridi and K. Ali and P. Magnusson},
  volume = 	 {966},
  series = 	 {LNCS},
  year = 	 {1995},
  publisher =    {Springer Verlag},
  address = 	 {Stockholm, Sweden},
  month = 	 {August},
  pages =        {15--26}
}

© Springer Verlag