This year’s general chair of Design Automation and Test in Europe is Donatella Sciuto, a full professor at the Politecnico d iMilano in Milan, Italy. She received her Laurea in Electronic Engineering from the Politecnico di Milano in 1984 and her PhD in Electrical and Computer Engineering in 1988 from the University of Colorado, Boulder. She also holds an MBA from the Scuola di Direzione Aziendale, Bocconi University. Her research interests include embedded systems design methodologies and architectures.
We spoke to Prof. Sciuto shortly after the DATE 2008 program was finalized. This is an extended version of this interview that originally appeared in EDA Tech Forum journal.
EDA Tech Forum: Your research covers areas such as embedded systems methodologies, hardware-software co-design and the shift to multi-processor systems-on-chip (MPSoCs), so let’s start there. How would you sketch that landscape, particularly given that DATE is dedicating a day to ‘dependable’ embedded systems?
Prof. Sciuto: We are now in the programmable platforms era. Some of the main problems facing EDA lie in finding the tools to build these platforms and in creating the software environments from which to program them. These problems are becoming increasingly relevant with the advent of ‘many-core multi-processors’. And these chips exist today – for example, the Cisco Systems Metrochip has 192 processing elements.
Meanwhile, applications and platforms are undergoing this merger of general computing, wireless connectivity and consumer electronics. That again increases the complexity of managing the design and programming tasks.
Our successes today have been only partial, dealing with single aspects of a growing problem and concentrating mostly on the development of programmable platforms. Major issues remain with the software-programming environment and in finding how to most effectively exploit these platforms.
In that context, there has been ongoing research on dependability for many years but its scope has been limited to specific mission-critical applications. The problem is that you need to be able to develop flexible, reconfigurable platforms, which feature that dependability. And you must be able to manage them. One of the goals for such platforms is to increase their lifetimes, even in the presence of changing requirements and additional features, to reduce development costs.
DATE this year returns to the ICM in Munich
So where do the various vendors stand in this environment? What role do they need to play?
EDA is still hardware-centric but must move towards a hardware-software system design focus. There has been much talk about the transition to ESL, but progress has been quite slow due to difficulty in providing generally applicable methodologies, rather than ones tailored to specific applications.
Multicore programming remains a largely unsolved problem and this creates what [EDA analyst] Gary Smith has identified as the software crisis. Since most applications are developed through sequential programming, one short-to-medium term solution might be tools that support parallelization and flexible system-level platform simulation, allowing the analysis of concurrent software execution in terms of performance and validation.
But remember – both the embedded software tools vendors and the RTOS vendors are themselves unsure about how to deal with multicore SoCs, so companies still tackle the problem with a mix of experienced programmers and internally developed tools.
Hardware-software co-design keeps cropping up as an objective, and we are again hearing a lot about it now because of the whole MPSoC trend.
I first began research into embedded systems design methodologies – and in particular, into concurrent hardware/software co-design – with the R&D staff of a telecoms company in the early 1990s. Most of the challenges we faced then are still here. They have just scaled up with the increasing complexity of the hardware platforms, including now MPSoC. They included – and still include – task partitioning, memory hierarchy, data communication and tasks interactions, bus structures or network-on-chip structures, and estimating and optimizing power consumption.
As one example, hardware/software task partitioning starting from the basis of a sequential application is still problematic. It has increased in complexity because of the move to multi-core architectures and the availability today of reconfigurable hardware components that accelerate execution.
The approaches adopted by system design companies are still very conservative but there is an interest in funding research that takes a broader view. Most major design companies here [in Europe] are involved in European-financed projects on these topics, including how EDA tools support the development of MPSoC. However, I don’t think the major EDA players will follow them into this area all that soon because of the difficulties they face in getting a return on their R&D investments in the short- or even medium-term.
So what is driving research from a European point of view?
New methodologies for embedded system design are valued because there is an understanding that they will directly impact on European competitiveness. In the next five years, embedded electronics components’ share in the value of a final product is expected to reach 22% in industrial automation, 33% in health and medical equipment, 37% in telecoms, and 41% in consumer electronics and intelligent homes. In the automotive sector, 20% of the value of a car today is attributed to embedded electronics; that is expected to rise to between 35% and 40% by 2015. More than 600,000 new jobs will be created in Europe in automotive embedded systems alone.
What do you see as the other priority areas for research here?
Other big issues that relate to managing design complexity today are time-to-market for high-volume applications, shortening design cycles, uncertainty over power and performance, and designs with hundreds of heterogeneous cores. I think the drivers behind the semiconductor business are also set to change, from consumer applications today to transportation, healthcare and clean technology during the next ten years.
How then would you describe the current state-of-play in European EDA research?
Academic research into EDA is Europe is tied to applications and system design, in particular for the telecoms and the automotive markets. In my opinion, semiconductor companies have reduced their investments in academic EDA research during the past ten years, focusing more on technology and less on EDA. However, together with the European Union, they have recently recognized the importance of EDA research by establishing two ‘joint technological platforms’ to sustain work in the areas of system design and microelectronics, ARTEMIS and ENIAC.
ARTEMIS aims to reinforce Europe’s leading position in embedded systems by accelerating the pace of innovation and countering any possible fragmentation of the R&D effort across the private and public sectors. It has done this by setting a common and compelling Strategic Research Agenda (SRA) which has the goal of nurturing an innovation eco-system based around world-class centers of excellence, and of gathering together systems integrators, technology providers, SMEs and academia to give them the capacity to rapidly turn research achievements into innovative products, processes and services for the global market. It aims to promote embedded systems as a respected part of the higher education curriculum, and finally, will ensure that its strategic objectives are properly coordinated. Ultimately, it provides the backbone of a true European Research Area for Embedded Systems.
ARTEMISIA is the next step in the implementation of the SRA for ARTEMIS, the European Technology Platform for Advanced Research and Technology for Embedded Intelligence and Systems that was launched in mid-2004. ARTEMISIA now extends that as the ‘ARTEMIS Industrial Association’. It was established in January 2007 with Philips Electronics, STMicroelectronics (ST), Thales, Nokia and DaimlerChrysler as its five founding fathers.
ENIAC is the European Nanoelectronics Initiative Advisory Council and it aims to provide a common European strategy for the full integration of nanoelectronics research at union-wide, regional and national levels, including relevant clusters in the intergovernmental EUREKA research initiative and private programs. Main members are ST, Infineon Technologies, NXP Semiconductors, Thales, Thomson Silicon Components, Robert Bosch, ASML, and SO.I.TEC.
Finally, let me say that the national investment strategies in research, in particular in EDA and embedded systems, differ significantly from country to country within Europe. As far as my own country, Italy, is concerned, there is no attention given to this type of research at government level and therefore no overall strategy and very little financing. Other countries in Europe have more interest in innovation and electronics research in general, and have national programs to foster growth in innovation.
How did paper submissions for this year’s DATE reflect European design priorities?
Submissions in the Embedded Software track increased by 50%, and we now think that DATE can show that it has firmly established itself as a true electronic system design conference because the same theme was present in other areas: in particular, design verification and system simulation, system-level power estimation and optimization, reconfigurable computing and fault tolerant systems. Our Applications track has also expanded this year, especially in the area of secure and security systems.
And Applications is where you are again showcasing a range of invited case studies highlighting best practices in industrial design.
One of the reasons for that is to give more visibility into the most advanced system-level designs already being undertaken by major companies. Also, there was no prior tradition for European system designers to write papers for conferences, so we see this as a first step, a way of showing them that these types of presentation have a value for them and their companies. It also creates an open forum where delegates can discuss their problems and share solutions.
This year we have two sessions in the Applications track, one on industrial system designs in transportation and a second on industrial designs in IT. We have papers from Thales, Robert Bosch, NXP, ST and Infineon.
It also sounds as though system design is important in differentiating DATE From DAC, particularly given that the two conferences are quite close in the calendar.
Yes, but it also reflects the fact that the European industrial and research environments are distinctly different from their US equivalents. The semiconductor industry is today US-centric, though it will probably be more Asia-centric a decade from now; Europe is more system-centric. Therefore the US focus is more on nanotechnologies, but in Europe the focus is on embedded systems and more recently on MEMS [micro-electromechanical systems]. Still, it was originally a challenge and more recently has been a critical success factor for DATE that it does differentiate itself from DAC. It took time but I think that the Applications track and the Embedded Software track are distinguishing factors. However, we retain the Design Methods track where discussion of traditional EDA research has an important role.
DATE has also been building up its workshops and ‘special day’ events.
The special days and the Friday workshops were launched with the specific purpose of introducing new and relevant topics every year that could eventually become part of the mainstream conference. The workshops also give people working in a particular field more space than is available in the main seven parallel sessions.
A very successful workshop has been the Reconfigurable Systems workshop, which grows steadily every year. Starting last year, the SCOPES conference, dealing mainly with compiler issues in embedded systems, decided to co-locate with us as a workshop and it brings new people to DATE. This year, increasing interest in MEMS has led to the organization of another workshop, again allowing us to reach out to a new community.
Automotive, as you said, is vitally important to the European technology sector. What are the more detailed trends you see there and also how will DATE’s dedicated Automotive Day address them?
With an increasing amount of electronic going into its cars, the automotive industry faces many different challenges, although the two most important are probably managing complexity and reliability. Furthermore, the role of converging communication systems and their impact on the value of a vehicle has to be taken into account.
There is a perceived inefficiency in today’s automotive electronics systems as the value chain has been based on the principle of ‘one function, one subsystem’. This approach is leading to an untenable situation as the number of electronic control units (ECUs) reaches into the high tens. Furthermore, there is a definite trend towards a more efficient partitioning of the functions on top of the hardware architecture.
This is causing profound changes in the value chain and highlights the need for new design and development methodologies. In addition, the lack of an overall understanding of the interplay between the sub-systems and of the difficulties encountered in integrating very complex parts – while guaranteeing safety, correctness and timely behavior – makes system integration very difficult.
The special day provides a comprehensive analysis of the evolution of automotive architectures – including ECUs, sensors and communication standards – and discusses how new methods, tools and standards for interoperability and component-based design can deal with the increasing complexity of software systems and their need for reliability and guaranteed timely behavior. It also looks at how well existing standards – like AUTOSAR, FLEXRAY and model-based development – can support the development of safety and time-critical software.
Could you discuss the thinking behind inviting this year’s keynote speakers, Dominique Vernay of Thales and Giovanni De Micheli of the EPF Lausanne?
My main goal for the keynote presentations was to identify two topics that complemented each other and were of interest both for research and for development. Also, in the past we have had representatives from major semiconductor and system companies in Europe as keynote speakers and I thought this year it would have been interesting to see the different aspects of embedded system design from the perspective of a major company that focuses on mission-critical systems, such as Thales.
So I asked Dr Vernay, the CTO of the company, to talk about Thales’ vision and research priorities for embedded systems and illustrate them through various solutions. The idea of to focus on advanced high-performance embedded computing platforms; on middleware technologies; on software systems design and verification tools for safety and security; and on the emergence of open standards in these domains.
Another area that interests me and one that is getting more and more attention in Europe is e-Health. Knowing that Prof. De Micheli was putting together a large project in Switzerland on smart micro and nano systems, I discussed the possibility with him of a talk looking at the design of such systems for healthcare applications.
Smart micro/nano systems will be the building blocks for wearable and ambient systems that gather and integrate heterogeneous data in real time and operate and communicate in a wireless and ultra low power mode. These systems will foster a revolution in health and environmental management, improving security and the quality of life. At the same time, they will create a large market for components and systems, and provide a renewed perspective for electronic design and manufacturing companies.
Finally, do you have any personal reflections on DATE’s importance to you over the years?
DATE has been and still is a very fulfilling experience for me since it has allowed me to grow from a professional point of view through my interaction with the other members of its Executive Committee. I have learned about the different aspects of managing such a complex event from the perspectives of both the scientific part and the commercial part of the event. It’s been very interesting also to participate in discussions on trends and strategies that are used to make DATE every year a better event, to make it more able to respond and anticipate the requirements of our community.
Design Automation and Test in Europe takes place from March 10-14 at the ICM conference center in Munich, Germany. More information on the event and online registration can be found at www.date-conference.com.