High performance, autonomy, reliability and safety of engineered systems require precise, controlled, highly concentrated and dynamic interaction of physical processes among system components. Embedded information systems are enabling technology for implementing these interactions. The presence and fundamental importance of cross cutting physical constraints in embedded software such as dynamics, noise, synchronization and dependability seriously limit the usability of tools and methods currently available for the composition of software systems.

A promising approach to achieve composability of embedded software is to use model-based integration technology in conjunction with embedded software frameworks. Embedded software frameworks include reusable services for time, synchronization, communication, fault management, as well as other, domain dependent common functionalities. Composability is achieved by enabling designers to capture the interrelationship of physical processes and the behavior of embedded software in the form of integrated, multiple-view models and use these models directly to customize framework components.

Model-based composition and integration of embedded systems require significant advancement in the following key technology areas:

1. New modeling approaches and tools need to be developed that offer better models of concurrency, facilitate the management and propagation of constraints among modeling views and allow domain-specific modeling without compromising solid semantic foundations.
2. New technology for the design and implementation of model-based generators need to be developed. The role of model-based generators is to translate relevant aspects of domain-specific models to customization interfaces of framework components and to input languages of analysis tools. Precise mathematical modeling, composition and automatic synthesis of model-based generators represent significant challenges for the research community.
3. Coupling and merging different embedded software frameworks is a very difficult problem. The lack of technology supporting this task is one of the fundamental reasons for limited reuse opportunities and high cost in embedded software. Model-based composition and integration technology has the potential to solving the problems. Explicit modeling of design constraints, computational models and assumptions implemented by different frameworks will enable using these models for analyzing composability of services, generating interface code among frameworks and moving embedded systems designs across platforms.

Introductory Presentation - Janos Sztipanovits - DARPA

Submitted Briefings

• Bob Balzer - USC ISI
• Ted Bapty - ISIS-Vanderbilt U
• Don Batory - U of Texas-Austin
• Robert Bond - MIT Lincoln Labs
• William Hodges - Boeing
• Gabor Karsai - Vanderbilt U
• Bill Koenig & Steve Hary - AFRL-IFSC
• Bob Laddaga - MIT
• Edward Lee - UC Berkeley
• Oz Levia - Improv Systems, Inc.
• David Martinez - MIT Lincoln Labs
• John Mitchell - Stanford
• Praveen Murthy - Angeles
• Charles Pecheur - RIAC-NASAAMES
• Raj Rajkumar - TimeSys Corporation
• Jerzy Rozenblit - U of Arizona
• Paul Robertson - DOLL
• John Rushby - SRI
• William Scherlis - CMU
• Kang Shin - U of Michigan
• Victoria Stavridou - SRI

Direct comments concerning this workshop to: John Luca