Embedded Software: Opportunities and Challenges

EECS Joint Colloquium Distinguished Lecture Series

pic of Janos Sztipanovits

Dr. Janos Sztipanovits
DARPA-ITO

Wednesday, January 31, 2001
Hewlett Packard Auditorium, 306 Soda Hall
4:00-5:00 p.m.

Abstract:

Among the many new developments in information technology, the fusion of information processing with physical processes literally changes the physical world around us. From electric shavers to airplanes and from cars to factory robots, computers monitor and control our physical environment. Arguably, this is the most aggressively growing applications of IT in general, and it represents the single most important opportunity and challenge for DoD. Virtually all new weapon systems from F22 to NMD and from FCS to UCAV depend on embedded software technology. The challenges DoD faces in building software for embedded information systems is indicated by the following problems well known for practitioners:

Is there any essential difference between embedded software development and software development in general? While no one argues that embedded software is much harder to design, the source and effects of differences have not been investigated well in the past. We argue that the root cause of the difficulties is that the role of embedded software has shifted in the last decade. Today, embedded information processing is becoming the dominant system integrator and we do not have design technology aligned with this new role. This emerging integration role of software has resulted in the emergence of fundamentally new challenges specific to embedded software technology, namely:

  1. physicality,
  2. change, and
  3. variable structures.

"Physicality" means that embedded software must be composed to satisfy conflicting physical requirements, such as dynamics, reliability, noise, power constraints, etc. "Change" refers to the expectation that embedded software - being the glue that keep the platforms together - shall provide controlled flexibility to tolerate and manage changes in physical platforms. "Variable structure" is an emerging new requirement for networked embedded systems, whose structure dynamically changes during operation. The talk describes ITO's research agenda designed to address these challenges and to utilize the opportunities of embedded systems offer.



Biography:

Dr. Sztipanovits (Fellow, IEEE) is currently a program manager at DARPA-ITO. He is responsible for the management of the Autonomous Negotiating Teams, Model-Based Integration of Embedded Software and Networked Embedded Software Technology programs. He is an IPA from Vanderbilt University, where he is E. Bronson Ingram Distinguished Professor of Engineering and founding director of the Institute for Software Integrated Systems. Dr. Sztipanovits received the Cand. of Tech. Sci. degree from the Hungarian Academy of Sciences in 1980, the Distinguished Doctor degree with the Golden Ring award form the President of Hungary in 1982, and the National Prize in Hungary in 1985. His primary research interest is software and systems engineering issues of embedded information systems, structurally adaptive systems and model-integrated computing. His work has been applied and fielded in several major aerospace and manufacturing industry projects. He has published over 130 papers; he is the co-author of two books and seven international patents.


231cory@EECS.Berkeley.EDU