added a research item
- Albert Boulanger
- Richard Shapiro
- Glenn Abrett
Carl Hewitt and his group at WT [Hewitt 84] have coined the term "open system" for large. distributed systems of computation which are open-ended and nonstationary. This paper will redefine the term "open system" to include the notion of open system in thermodynamics. i.e. coupling with an external environment. In this revised definition. as it will be shown below. the computational outcome of an open system is tied to the environment in which it is embedded. The outcome of an open computation depends on the environment because the system as a whole is a dynamical system. The asynchronous nature of the distributed computation of the individual components means that the time evolution of each component of the computation is relative to the other components with which it couples. In one of the open system computations presented below. it will be shown that. because of the ever-present coupling at the level of hardware. interaction goes beyond the immediate software level of the application to all levels of software and hardware and potentially to an external environment. In a tightly synchronized parallel or serial computation the dynamics of such coupling is normally irrelevant to computation because temporal precedence is known. Normally, the nonlinear nature of a computation where logic and decision making is present can be ignored. Thus. we can forget that the implementation of logic in a transistor is its nonlinear saturation region and make use of the macroscopic nonlinear switching behavior of transistors to build a computer engine to manipulate numbers and symbols. The nonlinear aspect of computation rears its head in an open system because dynamics at all levels of hardware and software can affect computation. The dynamics of computation in a distributed system is inherently non-linear. In addition to the nonlinear aspect of computer logic. the time delays introduced in relaxing temporal precedence contribute to the dynamics. Two examples of work that I have done with open system computation will be presented. The first example illustrates some of the phenomenologies of the asynchronous time evolution of each component of a parallel computation. A formalism based on time-delay equations from the theory of dynamical systems will be suggested for understanding the behavior of this computation. The second example illustrates open system coupling. This paper will conclude with an outline of future investigations. Two appendices will be included. One is a selected survey of the dynamical behavior of open system computation that has appeared in the literature. The second appendix is a short tutorial on dynamical system theory.
A novel system for extracting information from stereotyped voice traffic is described. Off-the-air recordings of commercial air traffic control communications are interpreted in order to identify the flights present and determine the scenario (e.g., takeoff, landing) that they are following. The system combines algorithms from signal segmentation, speaker segregation, speech recognition, natural language parsing, and topic classification into a single system. Initial evaluation of the algorithm on data recorded at Dallas-Fort Worth airport yields performance of 68% detection of flights with 98% precision at an operating point where 76% of the flight identifications are correctly recognized. In tower recording containing both takeoff and landing scenarios, flights are correctly classified as takeoff or landing 94% of the time
Grand challenge computing needs are influencing network architecture. Furthermore, protocols from emerging high speed networking applications, such as multimedia, are leading to new ideas in modeling. In this paper, we describe a practicable and scalable methodology for large-scale collaborative distributed computing of such grand challenge models as global climate change.