| Computational paradigms in which time plays a central role in defining the problem to be solved are generally called real-time paradigms . Real time is a generic term, encompassing the multitude of aspects encountered today in the systems area as well as in theoretical studies. An algorithm that solves a problem in real time must be able to handle a stream of input data, arriving during the computation. Usually, there is also a precise deadline that must be met when the solution is obtained. These are the main features characterizing the real-time paradigms investigated in this thesis.; Well-established paradigms like data-accumulation and correcting algorithms are addressed in this thesis. The study of novel paradigms, like those introduced by reactive real-time systems , is also initiated. The results obtained herein for all these paradigms; using different models of parallel computation, as well as the practical aspects of the problems investigated and the increasing importance of real-time requirements in today's computations should help parallelism earn its rightful place in computer science and motivate further research. (Abstract shortened by UMI.)... |
