Ptolemy Project Objectives

The Ptolemy project studies modeling, simulation, and design of concurrent, real-time, embedded systems. The focus is on assembly of concurrent components. The key underlying principle in the project is the use of well-defined models of computation that govern the interaction between components. A major problem area being addressed is the use of heterogenous mixtures of models of computation. A software system called Ptolemy II is being constructed in Java.

The class of systems addressed by the project is sometimes called reactive systems. Reactive systems are those that interact with their environment at the speed of the environment. They are often embedded systems, and have been contrasted with interactive systems, which react with the environment at their own speed, and transformational systems, which process a body of input data to produce a body of output data. Reactive systems typically include elements of signal processing, communications, and real-time control. They are typically implemented with mixed technologies, possibly including embedded software, custom digital hardware, configurable hardware, analog circuits, microwave circuits, and microelectromechanical systems (MEMS).

A key principle in the Ptolemy project is the use of multiple models of computation in a hierarchical heterogeneous design environment. A premise in this work is that no single general purpose model of computation is likely to emerge in the near future that will deliver what designers need. Modeling the diverse implementation technologies and their interaction is not reasonable within a homogeneous environment. Consider embedded DSP software controlling a MEMS device, for example. The MEMS device requires physical modeling using differential equations, a modeling technique that is entirely inappropriate for the DSP software. Moreover, the ability to validate designs and to synthesize high-quality implementations from high-level abstract models are both compromised by generality in the modeling framework.

The project aims to develop techniques supporting heterogeneous modeling, including both formal "meta-models" and a software laboratory for experimenting with heterogeneous modeling. In this context, it has explored methods based on dataflow and process networks, discrete-event systems, synchronous/reactive languages, finite-state machines, and communicating sequential processes. It has made contributions ranging from fundamental semantics to synthesis of embedded software and custom hardware.

The approach of the Ptolemy project at this time includes:

  • Use of programming language concepts such as semantics, type theories, reflection, and concurrency theories in system-level design of electronic systems.
  • Focus on domain-specific modeling and design problems so the designer can focus on the problem, not the tools.
  • Emphasis on understanding of systems, which is promoted by visual representations, executable models, and verification.
  • Use of Java, design patterns, UML, and a modern software engineering practice adapted to the realities of academic research.

Ptolemy software provides the laboratory for the experimental side of the project. It supports the interaction of diverse models of computation by using the object-oriented principles of polymorphism and information hiding. For example, using Ptolemy software, a high-level dataflow model of a signal processing system can be connected to a hardware simulator that in turn may be connected to a discrete-event model of a communication network.

Since the Ptolemy Project began, numerous advances in semantics, design, simulation, and code generation have occurred. Results of the Ptolemy Project have included:

  • adaptation of modern type theories to block-diagram-level specification of systems.
  • timed extensions of communicating sequential processes (CSP) and process networks (PN).
  • a modular approach to synchronous/reactive design.
  • a mathematical framework for comparing models of computation and studying their interaction.
  • a bounded execution policy for process networks.
  • formalization of computational models for dataflow.
  • managing regularity in dataflow graphs using higher-order functions.
  • automating the design of multidimensional multirate systems.
  • simulating and scheduling multidimensional multirate systems.
  • simulating heterogeneous systems.
  • modular hierarchical finite state machines with various concurrency models.
  • synthesizing embedded software from dataflow graphs.
  • parallel scheduling techniques.
  • optimizing interprocessor communication in parallel implementations.
  • deriving fast algorithms for hardware/software partitioning of dataflow graphs.
  • managing tool invocations and data dependencies in design processes.
  • integrated heterogeneous design visualization.

The Ptolemy software environment has been used for a broad range of applications including signal processing, telecommunications, parallel processing, wireless communications, network design, investment management, modeling of optical communication systems, real-time systems, and hardware/software co-design. Ptolemy software has also been used as a laboratory for signal processing and communications courses. Currently Ptolemy software has hundreds of active users at various sites worldwide in industry, academia, and government.