Excerpt
Goals
Today's world literally runs on embedded computers. Virtually every field of endeavor in our modern society depends on embedded computers from manufacturing to transportation to medicine. The typical household is a computing eco-system that includes telephones, televisions, washing machines, ovens, and a host of other silicon-based fauna. Many, if not most, of these computing devices have timeliness requirements to their functionality, so that late action is often wrong action. Many embedded devices have the capacity to do great harm if they malfunction or fail.
Not only are more things being handled by embedded computing devices, but the scope, complexity, and criticality of the things being handled is increasing geometrically. Technological advances are crucial in order to keep up with the increasing demands on the developer of such systems. Gone are the days when the hardware complexity was the limiting factor in the development of electrical devices. Most companies involved in the manufacture of real-time and embedded systems have realized the truism of "the tail that wags the dog" and have begun seriously looking at ways to improve software productivity. These better ways to develop real-time and embedded systems are the source and soul of this book.
Doing Hard Time: Designing and Implementing Embedded Systems with UML focuses on model-based development of real-time and embedded systems using the Unified Modified Language (UML) and a risk-based iterative development lifecycle called ROPES. UML is a 3rd generation modeling language that rigorously defines the semantics of the object metamodel and provided a notation for capturing and communicating object structure and behavior. The UML became a standard modeling language in the OMG in late 1996, and the author remains heavily involved in its ongoing effort. This book is based upon the 1.3 revision of the UML standard.
Model-based development is crucial in today's high-complexity, short-development-cycle business environment. It is important to focus on the fundamental abstractions of the problem rather than on the low-level details of its implementation; to focus on "should the control rods be in the reactor core to avoid a meltdown?" rather than "should I jump on non-zero or carry?" By increasing the level of abstraction, it is possible to build more complex systems with fewer defects in less time--a winning combination for everyone concerned.
Because the UML is executable, it is possible to automatically generate executable systems from UML models. The importance of this goes well beyond simply saving the time and effort of hand-translating code from abstract models. It is an enabling technology, allowing the developer to rapidly move from the inception of a concept to the testing of that concept. This allows early risk reduction and encourages exploration of the solution space. Conceptual defects can be identified and fixed very early before many dependencies on the flawed concepts are created, resulting in higher-quality systems in less calendar time.
This book is meant to be a fusion of a number of subject domains almost universally left disjoint--real-time concepts such as timeliness and performance, object modeling, a rapid development process, and system safety. This unified approach allows the developer to follow simple and well-understood process steps culminating with the delivery of correct and timely embedded solutions.
There are very few books on using objects in real-time systems and even fewer that use the latest in object modeling languages--the UML. Virtually all object-oriented books focus primarily on business or data base application domains and do not mention real-time aspects at all. On the other hand, texts on real-time systems have largely ignored object-oriented methods. For the most part, such books fall into two primary camps: those that bypass methodological considerations altogether and focus solely on "bare metal" programming and those that are highly theoretical with little advice for actually implementing workable systems. Doing Hard Time is meant to bridge for these technologies, presenting the development of deployable real-time systems using the object semantics and notation of the UML. It does so in a tool-independent manner, even though it does use a particular tool to demonstrate the examples.
Audience
The book is oriented towards the practicing professional software developer and the computer science major, in the junior year or higher. The book could serve as an undergraduate or graduate level text, but the focus is on practical development rather than a theoretical introduction. A few equations are to be found in this book, but more theoretical and mathematical approaches are referenced where appropriate. The book assumes a reasonable proficiency in at least one programming language and at least a cursory exposure to the fundamental concepts of both object orientation and real-time systems.
Organization
This book is organized into 5 sections:
This section introduces the object semantics and notation of the UML, real-time systems, safety criticality, and development processes. This section deals with the various kinds of analysis, including the capturing of requirements with use cases, scenarios, and state machines, the identification of the key abstractions in a problem, and modeling the fundamental behavior. This section focuses on the addition of design level information, such as the concurrency model, creation of run-time artifacts (libraries, executables, and so on), mapping to physical architectures, the application of design patterns to optimize collaborations of objects, and modeling of algorithms. - Advanced Real-Time Modeling
This section discusses topics of interest in difficult or complex real-time and embedded applications. These topics include the determination of the schedulability of object models through mathematical analysis, the reification of general state machine solutions to commonly occurring behavioral problems into behavioral design patterns, and the structure and function of real-time frameworks. This book provides three appendices: - UML Notational Guide: A short guide to the notations used in the UML and in the book.
- Introduction to Rhapsody: An introduction to the UML visual programming tool provided in the accompanying CD-ROM.
- Introduction to TimeWiz: An introduction to the schedulability analysis tool provided in the accompanying CD-ROM.
CD-ROM
The CD-ROM provided with this book contains three kinds of things:
- The examples presented within the pages of the book
These models are given as Rhapsody projects and can be opened and manipulated with Rhapsody. The visual programming tool installation files. The schedulability analysis tool installation files. I believe (and hope) that the needs of both the student and professional developer will be addressed by this book, and it is in this spirit that I offer it.
Acknowledgments
I wish to express thanks to my reviewers who tried hard to keep me honest and on topic, and who, I think, more or less succeeded:
- Eran Gery, i-Logix
- Jim Collins, i-Logix
- Larry McAlister, ENSCO, Inc.
- Therese M. Douglass, Air Traffic Software Architecture, Inc.
- Gary Cernosek, Rational Software Corp.
I would also like to thank Neeraj Chandra and Gene Robinson of i-Logix for their support in allowing me to spend so much effort on this book, Doug Jensen of Mitre for his input on schedulability, Therese Douglass for her expertise in air traffic control systems, and the editorial team at Addison-Wesley, including Carter Shanklin, Krysia Bebick, and Maureen Willard, to name a few. Bruce Powel Douglass, Ph.D.
Deep, Dark Winter, (early) 1999
0201498375P04062001