EMC2 a Platform Project on Embedded ...

EMC2 a Platform Project on Embedded Microcontrollers in Applications of Mobility, Industry and the Internet of Things Werner Weber

Alfred Hoess

Frank Oppenheimer

BEX RDE RDF Infineon Technologies Munich, Germany

Ostbayerische Technische Hochschule Amberg-Weiden Amberg, Germany

Transportation Division OFFIS e.V. Oldenburg, Germany

Bernd Koppenhöfer

Bastijn Vissers

Bjørn Nordmoen

Computing Platforms for Sensors Airbus Defence and Space Ulm, Germany

Imaging Systems Philips Healthcare Best, The Netherlands

Dept. Oslo WesternGeco Oslo, Norway

Abstract—EMC2 stands for “Embedded Multi-Core Systems for Mixed Criticality Applications in Dynamic and Changeable Real-Time Environments”. The goals of this ‘Artemis’ R&D project are to find solutions for dynamic adaptability in open systems, provide handling of mixed criticality applications under real-time conditions, scalability and utmost flexibility, full scale deployment and management of integrated tool chains, through the entire lifecycle. The project has started on April 1st 2014 and will run for three years. Keywords—Mixed criticality, dynamic application, real-time applications, multitasking, multicore, embedded microcontroller, open system, integrated toolchain

I.

INTRODUCTION

In the past few decades, microelectronics has advanced very quickly reaching amazing capabilities at lowered cost levels. Today, these features are primarily exploited in consumer-oriented products. The huge computing power of modern smart phones is just one of many examples. These systems are quickly put together since the next technology generation is already waiting around the corner. If one of them shows a malfunction the error may be tolerated and a new execution attempt started. In some cases switching off and on again may be a reasonable solution. This (and similar) way(s) of handling errors may be feasible for consumer products. However, in the professional areas such as industrial production, automotive, avionics, and space having to fulfill real-time safety requirements this simplistic approach is absolutely forbidden. The same argument holds in systems with high data volume such as multi data center applications. It is the prime task of EMC2 to bring the two worlds of today’s embedded microcontroller use and professional applications together and develop methods for safe use of modern embedded multicore controllers [1]. II.

PROJECT GOALS

Two questions arise on our way towards achieving this goal:

1. In which elements of the value chain do errors likely arise? Answer: Due to the extremely fast development of hardware technologies, the ubiquitous danger arises of being surpassed by the next generation product. Thus, ‘Time-toMarket’ becomes the prime paradigm of product developments. Value chain elements such as Module/System/Software suffer from this extreme pace since they are inherently slower in development than process and technology developments. This often results in shortcomings in terms of reliability and safety in particular in applications in which these elements are unnegotiable. 2. What are the technical elements to accomplish this task in particular in the value chain elements previously highlighted? Can we categorize the technical elements, key to those kinds of application? Answer: In this context, we need to point out that modern microcontrollers allow multitasking on multicore systems while many safety related applications so far use the approach of one device per application in order to avoid interference of different tasks. This is definitely a costly approach not fully exploiting the capabilities of modern hardware. Indeed, the following elements are central to develop inherently safe and cost-efficient systems using modern microelectronics (cf. Fig. 1) • Handling of mixed criticality applications under real-time conditions in one piece of hardware (most important) •

Dynamic Adaptability in Open Systems

• Utilization of expensive system features only as Serviceon-Demand in order to reduce the overall system cost. • Full scale deployment and management of integrated tool chains, through the entire lifecycle • Handling of a variable number of control units in one system The project focuses on the industrialization of European research outcomes and builds on the results of several previous Artemis, European and National projects. It intends to contribute to the goal of maintaining Europe’s excellent position in embedded systems.

WP1-WP6 will be applied and evaluated in the dedicated use cases of the living labs. State of the art in Criticality: ¾Handle different tasks in separate hardware

EMC2 Goals: ¾Handle mixed critical applications on one single die

State of the art in Complexity and Dynamics: ¾known number of control units and applications ¾test before runtime ¾Static scheduling of tasks, no dynamic changes at runtime

Technology–related workpackages complement dedicated Living Labs from various application domains representing European major industry sectors.

EMC 2 Goals ¾variable number of control units and unknown applications possible ¾full range of dynamic changes possible

Fig. 1. Cyberphysical System: Criticality, Complexity and Dynamics in Embedded Systems.

III.

MARKET INNOVATION AND IMPACT

2

The EMC project expects to facilitate EU’s ability to deploy and use Embedded Systems across important European market sectors by addressing the following main challenges • Automotive: Real time requirements and mixed criticality in embedded systems which are the key innovation driver enabling >90% of the innovations. • Avionics: The acceleration of technology cycles, and cost of software development. • Space: Increase of performance / weight ratio, high reliability and long life time • Industrial control and factory automation: The key areas will be energy efficiency and sustainability. • Healthcare: Work flow efficiency, integration of diagnosis and treatment, quality assurance. • Internet of things: Processing the increased amount of data available, as well as safety and security issues IV.

THE EMC2 PROJECT ARCHITECTURE

The EMC² embedded system approach will force the breakthrough and deployment of multi-core technology in professional application domains – Avionics, Space, Automotive, Medical, Energy, and Industrial factory automation – where real-time and mixed-criticality are an issue. The project is organized in a matrix structure with horizontal and vertical activities closely linked to each other (cf. Fig. 2). • Horizontal activities: Technological work packages (WP1-WP6) will develop dedicated technologies required for the implementation of embedded, mixed-criticality multi-core systems. • Vertical activities: So-called living labs (WP7-WP12) include several demonstrators for mixed-criticality embedded systems aiming at the same application domain. Each task in a living lab represents a use case. Technologies developed in

Fig. 2. Workpackage related architecture of the EMC2 project

Technology developments strictly follow the requirements derived from the specifications defined in each application domain. Their results are then fed back to the development of demonstrators on application level (Fig. 2). As a joint effort of technology work packages and application innovation work packages EMC² provides a flexible microprocessor System on Chip architecture that can be tailored by middleware to the needs of a particular application domain. This will substantially reduce the non-recurring development costs and the time to market of new embedded system applications and significantly cut the recurring cost of the respective products. Driven by the application needs EMC² puts a strong focus on reconfigurability and adaptability. The project will provide reference designs and architectures that provide solutions to key technical challenges – such as networking, security, robustness, diagnosis, maintenance, integrated resource management, self-organization, and – most notably – dynamic adjustments in changing systems. The project has 99 ordinary partners and one associated partner. They come from 16 European countries with Austria, Germany, the Netherlands, Sweden and Spain being the strongest stake holders. The project has a budget of € 90 Mio with € 40 Mio funding. Such a huge project can only be organized with a strongly decentralized management structure. Only certain key functions such contractual issues, reporting, inter-WP relations and external representation are kept centralized while most of the technical developments are handled in smaller working groups along specific value chains.

EMC2 project work packages are designed with unique objectives, task descriptions, deliverables, milestones, time plan etc., fitting seamlessly into the overall EMC2 plan. In the following, examples are presented in application areas of Automotive, Avionics, Health Care and Industry. V.

AUTOMOTIVE

Today, automotive applications are designed to use one microcontroller per application to avoid interferences between different applications. In this architecture, typically 100 microcontrollers are integrated into each car. This is costly and inhibits further applications. By allowing the integration of mixed critical and real time applications onto a single hardware platform, costs can be reduced and new application appear feasible (Fig. 3).

implementing a Helicopter Terrain Awareness and Warning System (HTAWS) (Fig. 4). VII. HEALTH CARE Today, in certain diagnostic scans problems in data acquisition not directly visible to the operator in the examination room may prevent the radiologist to draw conclusions from the post-processed image. Rescans with recall of the patient to the examination room are thus often required when data are only available several hours later. To avoid these problems we will bring reconstruction and postprocessing into the examination room by combining multiple mixed-critical systems on hardware level. There are challenges to manage mixed-criticality. Such architecture may reduce operation time per patients and thus reduce hardware and maintenance cost (Fig. 5). VIII. INDUSTRY In the following seismic surveilling is discussed which requires handling of massive amounts of data (Fig. 6). The goal is to produce images of geological features and their structure below the surface of the earth by triggering seismic events from the towing ship. 8-14 so-called streamers are towed by the ship whereby the Streamer length is 10 - 14 km. 100 - 200 computers are in operation per streamer and 200,000 sensors are active in each one of them. Real-time processing of the data requires a throughput of 300Mbits/s per streamer. This leads to more than 2000 computers in the streamers and more than 200 computers in the ship. This requires a total compute power of 2 Tflops, data rates of 1-3Gbits/s and a disk capacity of more than 100 Tbytes. IX.

CONCLUSION

2

Fig. 3. Reduce Number of Control Units and thus save cost and increase performance

VI.

AVIONICS

In avionics the reliability of systems is assured by governmental authorities, which require certification of the application. Certification Authorities (EASA, FAA) have concerns with respect to the usage of state-of-the-art multi-core processors in avionics since they fear unexpected interferences of applications. Thus strict separation of the cores must to be demonstrated. In EMC2 new monitoring methods are investigated to detect and prevent unwanted interference between cores whereby the methods exploit the chip manufacturer’s built-in test&debug functions. The effectiveness of the methods will be shown on a demonstrator

EMC brings together key European players from industry and academia. As an outcome of the project we expect the formation of new smaller consortia conducting new productoriented and funding-related projects in specific areas. EMC2 being a large platform provides the ideal basis for such development catalyzing these new groups to the benefit of European industry and research. ACKNOWLEDGMENT The research leading to these results receives funding from the ARTEMIS Joint Undertaking under grant agreement n° 621429 and from 16 national funding organizations. REFERENCES [1]

EMC2 public Webpage: http://www.artemis-emc2.eu/

Fig. 4. The helicopter terrain awareness and warning system supports pilots flying at night in changing weather with poor visibility in rough terrain or at low altitudes to prevent avoidable collisions with ground or obstacles.

Fig. 5. On the spot data reconstruction and post-processing of the scan data.

Fig. 6. Seismic Towing Configuration with numerous streamers towed by a ship taking data from seismic events triggered by the ship