Distributed real-time embedded systems: Recent advances, future trends and their impact on manufacturing plant control
ABSTRACT Real-time and embedded systems have historically been small scale. However, advances in microelectronics and software now allow embedded systems to be composed of a large set of processing elements, and the trend is towards significant enhanced functionality, complexity, and scalability, since those systems are increasingly being connected by wired and wireless networks to create large-scale distributed real-time embedded systems (DRES). Such embedded computing and information technologies have become at the same time an enabler for future manufacturing enterprises as well as a transformer of organizations and markets. This paper discusses opportunities for using recent advances in the DRES area in the deployment of intelligent, adaptive, and reconfigurable manufacturing plant control architectures.
- SourceAvailable from: Reiner Hartenstein
Conference Proceeding: A decade of reconfigurable computing: a visionary retrospective[show abstract] [hide abstract]
ABSTRACT: The paper surveys a decade of R&D on coarse grain reconfigurable hardware and related CAD, points out why this emerging discipline is heading toward a dichotomy of computing science, and advocates the introduction of a new soft machine paradigm to replace CAD by compilationDesign, Automation and Test in Europe, 2001. Conference and Exhibition 2001. Proceedings; 02/2001
- 01/1999; Coronet Books Hodder & Stoughton., ISBN: 978-0-8050-5874-1
- [show abstract] [hide abstract]
ABSTRACT: Open knowledge economy in intelligent industrial automation (OOONEIDA) is a new initiative for enabling decentralized, reconfigurable industrial control and automation in discrete manufacturing and continuous process systems. The goal of the OOONEIDA project is the creation of the technological infrastructure for a new, open knowledge economy for automation components and automated industrial products. This will be done by further development of the concept of reusable portable software modules (function blocks) and by their application in the time- and cost-effective specification, design, validation, realization, and deployment of intelligent mechatronic components in distributed industrial automation and control systems.IEEE Transactions on Industrial Informatics 01/2005; 1(1):4-17. · 3.38 Impact Factor
Distributed real-time embedded systems: Recent advances, future
trends and their impact on manufacturing plant control
Carlos Eduardo Pereira*, Luigi Carro
Electrical Engineering Department, Universidade Federal do Rio Grande do Sul (UFRGS), Brazil
Received 15 December 2006; accepted 16 February 2007
Real-time andembeddedsystemshavehistoricallybeensmall scale.However,advances inmicroelectronics andsoftwarenowallowembedded
systems to be composed of a large set of processing elements, and the trend is towards significant enhanced functionality, complexity, and
scalability, since those systems are increasingly being connected by wired and wireless networks to create large-scale distributed real-time
embedded systems (DRES). Such embedded computing and information technologies have become at the same time an enabler for future
DRES area in the deployment of intelligent, adaptive, and reconfigurable manufacturing plant control architectures.
# 2007 Elsevier Ltd. All rights reserved.
Keywords: Manufacturing systems; Real-time systems; Embedded systems
Market demands for innovative, high quality products,
aggressive competition at a global scale, increasing productiv-
ity through highly optimized production processes, and
environmental/societal pressures are some of the challenges
faced by the manufacturing industry today. Rapid changes in
process technology demand production systems that are
themselves easily upgradeable, and into which new technol-
ogies and new functions can be readily integrated (Mehrabi,
Ulsoy, & Koren, 2000). This situation has created the need for
novel manufacturing control systems that are able to manage
production change and disturbances, both effectively and
efficiently (Van Brussel, Wyns, Valckenaers, Bongaerts, &
Peeters, 1998), and has lead to the creation of concepts such as
‘‘flexible manufacturing’’ (Draper, 1984), ‘‘holonic manufac-
Nagel, & Preiss, 1995), and reconfigurable manufacturing’’
(Koren & Ulsoy, 1997; Mehrabi et al., 2000). All these
approaches aim to incorporate increased levels of flexibility,
reconfigurability, and intelligence into manufacturing systems
in order to meet highly dynamically marked demands. Molina
et al. (2005) presents a good overview on the historical
perspective and on key research issues in developing next-
generation manufacturing systems.
At the same time, an explosive growth in computer,
communication, and information technologies has been
experienced and manufacturing plants have also been affected
by this ‘‘pervasive and ubiquitous computing’’ era. The
manufacturing enterprise is intensively deploying a host of
hardware/software automation/information technologies in
order to face the changing societal environment pulled by
the increasing customization of both goods and services as
desired by customers (Morel et al., 2005). As the costs of
embedding computing becomes negligible compared to the
actual costof goods,thereisatrend ofincorporating computing
and communication capabilities in consumer products, and also
in manufacturing equipments. The intelligent manufacturing
field has been estimated to be larger than s300 million and
growing rapidly (Filos, 2004).
Such embedded computing and information technologies
have become at the same time an enabler for future
manufacturing enterprises as well as a transformer of organiza-
that has come to dominate the manufacturing floor (Johnson &
Annual Reviews in Control 31 (2007) 81–92
* Corresponding author.
1367-5788/$ – see front matter # 2007 Elsevier Ltd. All rights reserved.
Dausch, 2006). A great variety of the so-called e-Work and e-
Manufacturing activities (Nof, 2004) – such as virtual
manufacturing, augmented reality, intelligent maintenance,
intelligent supply, e-logistics – are now feasible and are being
adopted by several companies.
However, as pointed out in Morel et al. (2005), ‘‘only a form
of technical intelligence that goes beyond simple data through
information to knowledge and is embedded into manufacturing
systems components and within the products themselves will
play a prominent role as the pivotal technology that makes it
possible to meet agility/reconfigurability inmanufacturing over
flexibility and reactivity’’.
The INCOM Symposium series is the main technical event
supported by the IFAC Technical Committee 5.1 on Manu-
facturingPlant Control.INCOM aims todiscuss the application
of automation, information and communication technologies in
the control of the manufacturing plant and the entire supply
chain within the e-enterprise. While the topics discussed at
INCOM embrace all layers of the automation pyramid, from
low-level (sensors/actuators/industrial controllers), through
manufacturing execution systems to high-level e-enterprise
operations (virtual enterprises, supply chain, etc.), this paper
mainly focuses on the lower levels, the manufacturing plant
control or shop-floor level. At this level, there is a clear trend
towards distributed automation architectures, on which auto-
mation devices with local processing capabilities are inter-
(Mahalik, 2003). This distributed manufacturing automation
architecture heavily relies on an underling architecture
composed by the so-called distributed real-time and embedded
systems (DRES), ‘‘distributed’’ in the sense that devices/
machines are physically dispersed, but usually have to
exchange information in order to synchronize their operations.
The real-time characteristic is due to the fact that the
correctness of the system depends not only on the logical
results, but also on the time at which these results are produced.
While the areas of DRES and ‘‘intelligent, adaptive and
reconfigurable manufacturing’’ have usually been developed
apart, there is an increasing synergy among them, since recent
advances in DRES are enabling technologies for the develop-
ment of future intelligent manufacturing systems.
This paper discusses:
? Requirements imposed by industrial applications to the
computational entities embedded in sensors, actuators,
controllers, automated guided vehicles, industrial robots, etc.
? Recent advances on real-time distributed and embedded
computing elements and how these shall impact the
manufacturing plant control area.
? Development methodologies which aim to deploy DRES for
This paper is divided as follows: Section 2 describes some
intelligent manufacturing concepts that are enabled by DRES.
Section 3 lists requirements that are imposed on DRES to allow
their use in industrial applications, while Section 4 gives an
overview on recent advances and trends in DRES. In Section 5
some methodologies to develop DRES for industrial applica-
tions are discussed and also describes the SEEP methodology.
In order to illustrate the benefits that can be achieved by design
space exploration when deploying DRES for industrial
applications, two case studies are presented in Section 6.
Finally, Section 7 presents concluding remarks.
2. Examples of intelligent manufacturing areas enabled
by embedded computing systems
This section gives an overview on examples of intelligent
manufacturing concepts/methodologies which can benefit from
advances in DRES.
2.1. Agent-based manufacturing systems
The main reason for considering the application of multi-
agent systems to industrial applications is that other
technologies have not been able to meet all requirements
demanded by modern industrial automation systems: effective
enterprise integration, handling of cooperative and agile
processes, scalability, distribution, fault tolerance, interoper-
ability, among others. In many industrial scenarios, conven-
inadequate – especially under conditions of disruption and
long-term change – to cope with the high degree of complexity
and practical requirements for robustness, generality and
reconfigurability in manufacturing plant control as well as in
production management, planning and scheduling (Mar ˇı ´k &
These issues have led to the development of a new class of
approaches to manufacturing and supply chain decision
making. These approaches employ multi-agent systems
(MAS), which consist of a set of autonomous, intelligent,
their decision making to reach a higher-level or global goal. In
MAS groups of agents are organized according to specific,
precisely defined principles of community organization and
operation (such as messages and negotiation protocols), being
supported by an adequate agent platform (Luck, McBurney, &
Preist, 2001; Mar ˇı ´k & Laz ˇansky ´, 2004). Industrial applications
of agent-based technologies can be found mainly at three
different levels (Mar ˇı ´k & McFarlane, 2005): (i) real-time
manufacturing control; (ii) production management level
(encompassing problems such as planning, scheduling, orders
preprocessing, etc.); and (iii) virtual enterprises (VE) that will
integrate manufacturing, sales networks, suppliers, distribution
As long as completely different features and capacities are
required for each level, different agent’s concepts are applied.
For instance, at the real-time manufacturing control, the use of
agents with a simple reactive behaviour rather than deliberative
behaviour based on complex models and strongly proactive
strategies is prevalent. At this level, agents generally have one-
more detailed discussion on the timing requirements imposed
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 82
on MAS at this level as well as on how existing approaches
handle the almost contradictory goals of being adaptive and
flexible, while also presenting a deterministic temporal
behaviour is presented in (Pereira & Mitidieri, 1999).
As higher the level of the manufacturing automation
pyramid, real-time requirements tend to become softer and
knowledge intensive decision processes are performed by
in Shen and Norrie (1999), Mar ˇı ´k and McFarlane (2005),
2.2. Holonic Manufacturing Systems (HMS)
As originally introduced by Koestler (1989), the word
‘‘holon’’ aims to describe the hybrid nature of sub-wholes/parts
in real-life systems: holons simultaneously are self-contained
wholes to their subordinated parts and dependent parts when
seen from the inverse direction. From this viewpoint, the whole
factory can be considered as a holon being composed of other
different kinds of holons such as those representing physical
objects like machines, automatic guided vehicles, conveyor
belts, pumps, valves, and even products as well as non-physical
entities like customer orders, production plans and so on.
Holonic Manufacturing Systems (HMS) are manufacturing
systems structured as a set of holons, which are autonomous
and cooperative units (see Deen, 2003; Van Brussel, 1994; Van
Leeuwen & Norrie, 1997). While this definition is somewhat
similar to the one presented in agent-based industrial
applications, according to Mar ˇı ´k and Laz ˇansky ´ (2004) holons
can be considered as specific reactive agents which are strongly
connected with the physical level devices, operate in hard real-
time and can be organized into a ‘‘holarchical’’ structures
(hierarchical or heterachical). In Mar ˇı ´k and Pechoucek (2001),
mutual impacts of holons and agent concepts are described.
In Van Brussel et al. (1998), a reference architecture for
Holonic Manufacturing Systemsnamed PROSA ispresented. It
consists of three types of basic holons: resource, product and
order holons. PROSA makes use of object-oriented concepts
such as aggregation and specialization to structure the holons.
Holons, as cooperative units, exchange information about the
manufacturing system. Product holons and resource holons
communicate process knowledge, product holons and order
holons exchange production knowledge, and resource holons
and order holons share process execution knowledge.
The IEC 61499 standard for the application of function
blocks in distributed industrial-process measurement and
control systems has been developed by the Holonic Manu-
facturing Systems (HMS) consortium for the holonic real-time
control (Christensen, 1994). Function blocks are used as a
container of application algorithms and services wrapped with
execution control, promoting reuse of existing algorithms and
modular development of manufacturing systems (Fletcher &
Brennan, 2001). Additionally, IEC 61499 also includes models
for distribution, communication, and services.
As it is discussed later, distributed real-time embedded
architectures provide a very interesting infrastructure to deploy
such Holonic Manufacturing Systems.
2.3. Intelligent maintenance systems
Taking into consideration the fact that machines usually do
not suddenly fail, butrather go through a measurable process of
degradation before they fail, the basic idea of intelligent
maintenance systems or intelligent prognostics systems is to
use information provided by sensors and computerized
components embedded on the equipments, and apply algo-
rithms for health estimation and failure prediction.
The fundamental basis for such intelligent maintenance
systems (see Fig. 1) are the so-called ‘‘infotronic technologies’’
(Lee, Qiu, Ni, & Ad Djurdjanovic, 2004), which ‘‘transform the
the use of intertwined embedded informatics and electronic
intelligence in a networked and tether-free environment and
Fig. 1. Intelligent maintenance systems (Lee et al., 2004).
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 83
enables products and systems to intelligently monitor, predict
and optimize their performance and ultimately to perform self-
maintenance activities autonomously’’.
Embedded computing components such as embedded
sensors, intelligent actuators and processing elements for local
smart decision devices play a fundamental role in the
development of such intelligent maintenance systems. As it
is presented later in the paper, recent advances in areas such
Systems-on-Chip (SoC), reconfigurable hardware, etc., are
enabling technologies for the development of embedded
systems for intelligent prognostics.
2.4. Product lifecycle management using product
embedded information devices
Product lifecycle management (PLM) can be described as a
strategic business approach that applies a consistent set of
business solutions in support of the collaborative creation,
management, dissemination, and use of product definition
of-life—integrating people, processes, business systems, and
information (Kiritsis, 2004). PLM consolidates diverse busi-
ness activities that create, modify and use data to support all
phases of a product’s lifecycle from ‘‘begin-of-life’’ (design,
production), middle-of-life (use, maintenance), and end-of-life
A key component in modern PLM systems is the concept of
‘‘smart items’’, physical objects that are equipped with
embedded computing units to enable close computing of the
real world to backend information systems. Such embedded
computing units are the so-called PEIDs (product embedded
information devices), which usually contain RFID tags, sensor
nodes, embedded PCs or similar devices (Anke & Neugebauer,
2006). This technology will allow producers to dramatically
responsibility as producers of environmental friendly and
PROMISE is an international project on product lifecycle
management and information tracking using smart embedded
systems that is being carried on within the scope of the 6th
Framework Program of the European Union and as an endorsed
IMS project (http://www.promise.no/). PROMISE’s main goal
is to allow information flow management to go beyond the
customer, to close the product lifecycle information loops, and
to enable the seamless e-Transformation of Product Lifecycle
Information to Knowledge (see Fig. 2).
As depicted in Fig. 2, PROMISE concepts rely on a DRES
infrastructure to track objects’ identity, status, and location and
3. Requirements to embedded computing systems for
manufacturing automation applications
The great majority of embedded systems currently being
developed and deployed are for use in mass markets such as
consumer electronics and their use in industrial applications are
still a small but increasing percentage. Similarly, the rapid
progress in COTS software for mainstream business systems
has not yet become as broadly available for DRES.
However, as presented in the last sections, there are several
industrial applications. However, in order to be applicable
to industrial applications, DRES have to meet following
Fig. 2. PROMISE concept (from www.promise.no).
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 84
? Dependability: it is usually defined as that property of a
computer system such that reliance can justifiably be placed
on the service it delivers. Both for economical (for instance,
high costs of breakdown time) as well as for safety reasons,
dependability is a key concept that must be supported by
DRES when applied to industrial applications.
? Real-time communication: most manufacturing systems are
physically distributed over a plant site, so that that their
embedded system components will also be physically apart.
In order to be able to interact and synchronize while meeting
stringent timing requirements, real-time industrial commu-
nication protocols must be employed (Mahalik, 2003;
Neumann, 2007), so that a timely communication occurs.
and industrial processes will exhibit much higher degrees
of physical reconfigurability in order to accommodate
frequent changes in product mix and volume, as well as
due to frequent introduction of new product types and
manufacturing technology. In addition, rapid reconfiguration
will be used much more frequently to recover from machine
and process faults with minimal loss of production. In all
these cases, the control system (hardware and software) must
be quickly reconfigured, and for the most part automatically
so, in order not to become a bottleneck to agility.
? Modularity: in order to support the above-mentioned
requirements of adaptability, DRES must be constructed in
a modular way. Modularity also affects serviceability and
recyclability in terms of disassembly, separation, repair, and
reprocessing (Ishii, 1998).
? Openness: a system is defined as open when the implementa-
tions of its components conform to an (non-proprietary)
interface specification such that upgrading and customization
of the system as well as integration of new components is
possible (Mehrabi et al., 2000).
? Location transparency: communication among different
nodes should use names that are not dependent on user’s
or resource’s location. This becomes particularly important
when mobile devices/equipment, such as AGVs, are used.
? Autonomous behaviour: as discussed in Section 2, a key issue
in intelligent manufacturing systems is the capability of
manufacturing devices in autonomously making decisions
and local data process capabilities.
? Security: embedded computing systems need to access, store,
manipulate, or communicate sensitive information and
frequently need to operate in physically insecure environ-
ments. In industrial applications it is mandatory to have a
process to prevent and detect unauthorised use of services
(Ravi, Raghunathan, Kocher, & Hattangady, 2004) presents a
good overview on design challenges for deploying secure
4. Distributed real-time and embedded systems (DRES)
4.1. Embedded systems hardware and SOCs
Embedded systems are defined as computational resources
that take part in bigger systems. They may vary from ABS
(antilock brakingsystems)control ina car tothe portablephone
or home set-top box, and nowadays they include almost all
systems that are not desktop computers. Helped by the
technology advances allowed by Moore’s law, that states that
every 18 months the number of transistors on a single die
doubles, embedded systems complexity has increased follow-
ing the same pace. Semiconductor companies are now
fabricating Systems-on-Chip, or SoCs (Bergamaschi et al.,
2001), complex devices that include one or more processors,
plus memories and communication resources in a single die, at
a cost in the range of few dollars. Portable phones are classical
examples of products that benefit from such technology.
This way, the availability of fast operating devices with
embedded memories and other resources can be almost taken
for granted. For example, the same seamless connection that
today benefits persons while using a wireless link inside
buildings or airports can be transported to the manufacturing
scenario. The low cost of RF devices can provide new access
means for plant observation and control, without wiring costs
and accessibility problems.
Another key issue of current embedded devices is their
adaptability. Thanks to the huge processing power available,
and the reduced costs of embedded flash memories, these SOCs
can be programmed using a regular C or C++ compiler, thus
allowing fast adaptability to any task. Moreover, embedded
operating systems are also available for these devices, making
the porting of new applications an easier task, when compared
to the way things should be done in case of dedicated hardware
and software platforms.
The last technological evolution to enter the scenario is the
possibility of hardware reconfiguration itself (Hartenstein,
2001). Reconfigurable hardware allows for another level of
programmability that could boost performance at a fraction of
the energy spent by a microprocessor.
Besides regular programmable devices like FPGAs (e.g.
commercial trend is found nowadays, concerning the reconfi-
gurations of processors themselves, either by changing the
instruction set or by adding extra hardware tightly connected to
the instruction set, like in the Coware and Tensilica approaches
(www.coware.com and www.tensilica.com). The advantage of
changing the processor concerns the huge amount of software
reusability allowed, while still providing extra performance
thanks to the dedicated instructions.
This technological scenario shows that embedded systems
can be as complex as required, and thanks to their mass
production, they can be available at a reasonable cost. From the
manufacturing and plant control point of view, these advances
might mean that the number of observation and control points
might increase, without extra wiring required and without huge
investments. Moreover, thanks to their high processing power,
that adaptability on the field can be deployed at a faster pace.
Hardware components in a SoC include one or several
processors, even from different types (microcontrollers, DSP,
RISC), memories, dedicated components for accelerating
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 85
critical tasks, and interfaces to various peripherals. Compo-
nents are connected by arbitrary communication networks,
which may range from a simple bus to hierarchical buses
connected by bridges to a complex network-on-chip (De
Micheli & Benini, 2002), as illustrated in Fig. 3.
The design of embedded systems is becoming largely
software-dominated, and is tightly coupled to a platform
(Densmore, Passerone, & Sangiovanni-Vincentelli, 2006).
Market perspectives indicate that up to 90% of the embedded
system design effort is now on the software part. Software
components in a SoC include several application software tasks
running on the processors, device drivers, and a real-time
operating system (RTOS) (Burns & Wellings, 1997) for each
available processor, to support basic services such as
scheduling and communication. An RTOS, besides usual
functions of an operating system, must also fulfill application
temporal requirements, such as task deadlines and frequency of
activation of periodic tasks. Temporal restrictions have impact
ontask scheduling algorithms and on responsetime of basic OS
services such as interrupts and context switching.
Major guidelines for the SoC design are the clear separation
between computation and communication (Rowson & Sangio-
vanni-Vincentelli, 1997) and between function and architecture
(Keutzer, Malik, Newton, Rabaey, & Sangiovanni-Vincentelli,
2000; Sangiovanni-Vincentelli & Martin, 2001). These
distinctions enforce modular design and promote the indepen-
dent design and evolution of three aspects of system design:
function, architecture, and communication.
As depicted in Fig. 4, the design of an embedded SoC starts
with the definition and validation of a high-level, pure
functional specification, which is not influenced by architec-
tural choices and does not consider how design requirements
(power, performance) may be fulfilled. Following a design
space exploration step, which is nowadays mainly a manual
functionality, and a mapping assigns functional blocks to
architectural ones. This mapping implements a hardware–
software partitioning, where some functions are mapped to
software tasks while other ones are mapped to dedicated
hardware blocks. This high-level architectural model abstracts
all low-level implementation details.
A performance evaluation of the system is then performed,
for this macro-architecture. Results of this estimation will be
used to guide the design space exploration, for instance
requiring modifications in the chosen macro-architecture and/
or functional mapping.
Communication refinement is now possible, where high-
level communications are mapped to particular mechanisms,
protocols and channels, thus allowing a more precise
performance evaluation. Depending on the chosen commu-
nication mechanisms, specialized components such as DMA
and interrupt controllers and bus arbiters must be inserted.
Some approaches implement the automatic generation of these
components (Lyonnard, Yoo, Baghdadi, & Jerraya, 2001;
O’Nils & Jantsch, 2001).
Hardware and software synthesis follow, usually resulting in
C/C++ code for the software part and an HDL description of a
cycle-and-pin accurate micro-architecture for the hardware
part. This synthesis is largely automated, and the combined C-
HDL description may be validated by conventional co-
simulation tools, such as CoCentric System Studio (Synopsys,
2003), from Synopsys, and Seamless CVE (Mentor, 2004),
from Mentor. As part of the software synthesis process, an
operating system may be required, to implement for instance
services for communication among tasks and for scheduling
tasks that are mapped to the same processor.
4.2. Middleware and programming languages
Real-time and embedded systems have historically been
relatively small scale. As discussed in the last section, recent
advances in microelectronic and software now allow embedded
systems to be composed of a large set of processing elements,
and the trend is towards significant increased functionality,
complexity, and scalability, since those systems are increas-
ingly being connected by wired and wireless networks to create
large-scale DRES. Additionally, the environment is generally
non-static, and the whole system must be robust enough in
order to operate under highly unpredictable and changeable
conditions. An important and challenging problem for DRE
systems is therefore adaptation of behaviour and reconfigura-
tion of resources to maintain the best possible application
Fig. 3. Typical SoC hardware architecture.
Fig. 4. A generic SoC design methodology.
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 86
performance in the face of changesin system load and available
resources (Schantz et al., 2006). Therefore, new techniques and
software solutions are required in order to properly handle the
increased system complexity.
Clearly, a changing environment requires extra adaptability
from the software and hardware resources. Not only one must
provide for these adaptations, but also one must take into
account the possibilities of exploiting the architecture of the
system itself. For example, as more intelligent nodes are
available at low cost, distributed processing starts to make a lot
of sense, since the cost of local processing is not only
energetically efficient, but it is important to notice that
bandwidth does not follow the advances of Moore’s law.
The need for autonomous and time critical behaviour in
manufacturing plant control demands a flexible distributed
system substrate that adapt robustly to dynamic changes in
application requirements and market/environment conditions.
This substrate is usually called ‘‘middleware’’ because they sit
‘‘in the middle’’ in a layer above operating systems and
networking SW/HW and below industry specific applications
(Bernstein, 1996). Schantz and Schmidt (2001) define
middleware as reusable systems software that functionally
bridges the gap between: (i) the end-to-end functional
requirements and mission doctrine of applications and (ii)
the lower-level underlying operation system and network stack
protocols. Middleware therefore provides capabilities whose
quality and quality of service (QoS) are critical to DRE
Similar to network protocols, middleware can also be
decomposed in several layers (Schmidt, 2002): from ‘‘host
infrastructure middleware’’ at lower level to ‘‘domain specific
middleware services’’ at higher level (below the ‘‘application
level’’). Each of these layers focuses on specific aspects, but all
have in common the idea of allowing the implementation of an
‘‘information utility’’, to which components such as manu-
facturing devices can be connected and are able to interact with
Some of the middleware provided functionalities are: (i) to
encapsulate and enhance native OS communication and
concurrency mechanisms to create portable and reusable
network programming components (connection management,
data transfer, parameter (de)marshalling, etc.), (ii) to minimize
hardware and software infrastructure dependencies, and (iii) to
allow management of processor, memory, and communication
Domain-specific middleware services are tailored to the
requirements ofaparticular domainand havethemostpotential
to increase the quality and decrease the cycle-time and efforts
that integrators require to develop a particular class of DRE
systems (Schmidt, 2002). For instance, for manufacturing plant
control applications, this middleware level could for instance,
support all holon types defined by the PROSA reference
architecture as discussed in Section 2. It is also important to
note that in applications such as manufacturing plant control,
middleware must allow functional and QoS-related properties
to be modified dependably, i.e. without compromising the
fulfillment of stringent timing requirements. That means, a key
aspect is to achieve a good balance for the trade-off
performance versus flexibility.
TAO (Schmidt, Levine, & Mungee, 1998) and QuO (Loyall
used for manufacturing plant control applications.
Language support for embedded software development
efforts are currently centred on the C and C++ programming
languages but Java is rapidly gaining momentum in this field.
Deviceswith embedded Javasuchascellularphones,PDAsand
pagers have grown from 176 million in 2001 to nearly 800
million in 2005. It has been predicted that at least 80% of
mobile phones will support Java by this year (Lawton, 2002). A
key concept in Java is that Java byte codes can be run on any
architecture for which a customized Java Virtual Machine
(JVM) is available. The JVM encapsulates all platform-specific
services, such as networking, file system operations, etc. The
idea of using Java for embedded industrial applications is not
new (see for instance (Atherton, 1998)) and with the advent of
the so-called Real-Time Specification for Java (RTSJ)
(Bollella, Gosling, & Benjamin, 2001), a very interesting
alternative to the development of DRES became available.
RTSJ incorporates several useful real-time constructs into Java,
allowing periodic activation of concurrent processes, timed
actions, handlers for asynchronous events, etc. Wehrmeister,
deployment of real-time embedded applications based on
4.3. Real-time communication protocols
Real-time communication protocols are an important
component in DRES for industrial applications in order to
ensure a safe and timely operation. Fiedlbus protocols such as
Profibus, Foundation Fieldbus, DeviceNet and CAN are
standardized (IEC, 2003), widely adopted and well established
at field/shop floor level (see for instance Mahalik, 2003).
Ongoing efforts in extensions to these industrial communica-
tion protocols have shifted from low level aspects (physical and
data link layer standards) to the definition of higher-level
automation objects, such as Profinet mechatronic objects
(Profibus, 2002) or CIP application layer objects (www.od-
At the same time, Ethernet has also been considered for use
in real-time applications, either in the industrial domain or in
large embedded systems. Attractive factors include wide
availability, high bandwidth and low cost. The use of
Ethernet-based communication protocols should also enable
an easy integration to realise the access to data invarious layers
of an enterprise information system. As already mentioned,
these different levels impose different requirements dictated by
the nature andtype ofinformation beingexchanged.Due tofact
that Ethernet was not originally developed to meet real-time
requirements and the medium access control protocol used –
CSMA/CD – may cause unbounded network access delays,
several proposals have been presented to adapt this protocol in
order to achieve real-time behaviour (Neumann, 2007;
Pedreiras & Almeida, 2005).
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 87
4.4. Future trends in DRES
The continuous trend towards smaller, more intelligent, and
more numerous devices is continuous and very soon also
embedded computing systems will be approaching limits of
human capability to develop, operate and maintain these
systems. It can be observed that objects are becoming
intelligent ‘‘things that think’’ (Gershenfeld, 1999) and while
this scenario will enable the realization of several high value e-
manufacturing and e-services, completely new design and
operation paradigms have to be developed. This has been the
motivation to the creation of areas such as autonomic
computing (Kephart & Chess, 2003) and organic computing
(DFG, 2004; Schmek, 2005). The idea is to develop
(embedded) computing systems that manage themselves given
self-optimization, self-healing and self-protection (the so-
called ‘‘self-x’’ properties). Those systems should have
sufficient degrees of freedom to allow a self-organized
behaviour which will adapt to dynamically changing require-
ments. The ability to deal with widely varying time and
resources demands while still delivering dependable and
adaptable services with guaranteed temporal qualities is a
key aspect for future DRES (Stankovic, 1996). Some examples
of autonomic and reconfigurable embedded real-time systems
and Go ¨tz, Rettberg, and Pereira (2005).
Clearly, the possibility of using low cost devices that can be
configured, either in software or in hardware to adapt its
characteristics to those better matching the underlying
environment is beneficial to the whole design process. The
major challenge, however, is how to provide engineers with
effective and productive tools to allow them to make these
complex systems in a timely manner. To develop the next
generation of open, modular, reconfigurable, maintainable, and
dependable manufacturing systems adequate methodologies
must be available. When dealing with complex industrial
automation applications, the definition of a good architecture is
of utmost importance. Aspects such as modularity, cohesion,
and coupling, which historically were relegated to a secondary
plan due to an overemphasis on systems performance, have a
major impact in installation, operation, maintenance, and
engineering costs. Object-oriented systems have important and
desirable architectural properties. They are composed of a
number of communicating and well-defined objects. Objects
with common characteristics and behaviours are organized into
classes. Class hierarchies can be built using inheritance
concepts. Objects also fit nicely with concurrence, since their
logical autonomy makes them a natural unity for concurrent
execution. That implies in a fruitful way of thinking, enabling
concurrent processes present in the real world to be expressed
in a natural and easily understandable way. Some examples of
object-oriented technologies for industrial applications are
described in the sequence.
SIMOO-RT (Becker & Pereira, 2002) is an object-oriented
framework to the development of real-time computer-based
systems (i.e. hardware and software) that are embedded in
devices used in flexible and adaptive industrial automation
systems. The approach is based on the concept of active objects,
which are autonomous and concurrent processing units, having
their own thread of control. Active objects are used to map the
structure and the desired behaviour of technical plant compo-
nents. The approach leads to a generic specification, which
preserves the semantics of the physical plant under automation.
systems: from requirements engineering, through hardware and
software design, and to implementation and validation.
DOME (Distributed Object Model Environment) is an event
driven, object-oriented distributed architecture on which
industrial are defined as a network of Automation Objects,
which can be assigned to different contexts on several
heterogenous nodes. The access to the process interface is
also encapsulated inside DOME objects, which can be treated
as a kind of service interface function block (SIFB) according
to IEC 61499 or as a Proxy-object. A DOME case study is
presented in Riedl, Diedrich, and Nauman (2006).
OONEIDA (Vyatkin, Christensen, & Lastra, 2005) is a
research and development initiative in the domain of
decentralized, agile industrial control and automation for both
discrete manufacturing and continuous process systems.
OONEIDA should enable all players in the automation value
creation chain to encapsulate their intellectual property into the
software components and to deploy these components into
intelligent devices, machines, systems, and automated fac-
tories, respectively. This will enable time- and cost-effective
specification, design, validation, realization, and deployment of
intelligent mechatronic components in distributed industrial
automation and control systems.
Ptolemy (http://ptolemy.eecs.berkeley.edu) is a design
methodology based on the description of complex behaviour
with the use of OO languages. The main goal of Ptolemy is to
raise the abstraction level on the design process of systems
composed of hardware and software components. The project
goal is to cover the modelling, simulation and design of
concurrent components. Several models of computation are
supported, and basically the task of the designer is to use a
heterogeneous mixture of several models to describe and
simulate a complex system.
SEEP is a design and verification methodology that allows
the development of DRES from high-level RT-UML models.
The SEEP project (SEEP, 2006) concerns the development of
embedded systems based on platforms. A platform is defined as
a combination of hardware and software resources, where the
customization of these resources is developed for a target
In the specification phase, the user might use UML
diagrams, and a Simulink frontend is being currently adapted.
From UML one can make the fist design exploration, by using
the developed library plus a tool that evaluates the costs, in
terms of processing speed, memory and power of the available
solutions (Brisolara, Becker, Carro, Wagner, & Pereira).
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 88
After this first design exploration step, SEEP supports two
different platforms, the Power-PC and a Java processor in
several organizations (pipeline, VLIW and with a reconfigur-
able array). Each of these processor versions has a different
trade-off regarding area, power dissipation and speed. Also,
each one can be synthesized only with a dedicated instruction
set, strongly tied to the problem it is trying to solve, so that
maximum efficiency can be achieved with software compat-
ibility. Examples of such architectures being deployed can be
found in Krapf and Carro (2003), Beck and Carro (2004, 2005),
and Silva et al. (2006).
6. Case studies
In order to better illustrate possibilities of design space
exploration when synthesizing embedded systems for manu-
facturing automation applications, two case studies are
6.1. Customizable RT-Java-based SOC for holonic/agent
The first case study deals with the deployment of a SOC for
the holonic manufacturing case study described in McFarlane
(2002), real-time manufacturing control tested which consists
of a flexible production cell containing a robot arm, a screwing
robot, a rotary table, a flipping unit, an item parts input and a
storage unit to store the assembled items. Different product
types are assembled according to product specification.
The proposed case study was implemented using the SEEP
methodology described in previous section. A RT-UML model
encompassing around 30 classes was created (due to
restrictions in paper’s length this diagram is not presented
here) and a SOC containing a customized RT-FemtoJava, an
API based on the Real-Time Specification for Java (RTSJ)
(Wehrmeister et al., 2004), and a real-time communication API
(Silva et al., 2006) was synthesized using the SASHIMI
synthesis tool (Ito, Carro, & Jacobi, 2001). This case study is
partially implemented, however all important functions are
present in the current system version. Table 1 depicts the
achieved optimization. The first column describes the required
footprint for a single node application, i.e. when all system
objects inhabit in one computing device. The optimized system
requires 88.28% less hardware resources when compared with
the same application running on a standard JVM and processor.
Considering that most Java applications need to be executed
reduction would be 96.27%, because most of the code included
in the non-optimized application would not be used. In the
distributed version, the application must include the real-time
communication API and the RTSJ-base API in each system
node. Even with this extra code the reduction remains about
88% to application code size and about 96% when compared to
the same application running on a commercial RTSJ-
6.2. Embedded prognostics SOC
As a second case study, let us consider the design of an
embedded prognostics SoC to be used in intelligent prognostics
or condition-based monitoring applications. The goal here is to
use intelligent embedded prognostics algorithms in order to
perform a continuous assessment and prediction of a product’s
degradation performance by extracting high-level information
– the so-called features – from sensory signals. As discussed in
Lee et al. (2004), examples of such algorithms include Kalman
filters, time-frequency based, time-series based and wavelet-
based systemanalysis,Autoregressive Moving-Average
Flexible assembly—HW/SW optimization
Single nodeDistributed nodes
Robot armScrewing robot Assembly
Original code size
Total + RTSJ JVMa
Synthesized code size
Used API classes
App. code size
App. + RTSJ JVM
App. code size
App. + RTSJ JVM
aThe RTSJ JVM size was based on a commercial embedded JVM.
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–9289
(ARMA) analysis, Hidden Markov Models, sensor fusion
techniques, etc. For instance the IMS Center for Intelligent
Maintenance Systems in U.S. (IMS Center, 2006) has
developed a set of around 20 prognostic tools, based on
different algorithms, which are used for feature extraction,
performance assessment, diagnostics and prognostics and have
been successfully applied to different industrial testbeds. More
than one algorithm can be applied in a particular application
and obtained results can be fused using some averaging
technique in order to increase the robustness and quality of
obtained confidence values.
An embedded system to implement such an embedded
prognostic device includes: I/O modules for data acquisition of
digital and analog signals, communication modules, both for
industrial communication protocols as well as for Internet
connectivity (for instance, with an embedded Web server to
allow remote access and configuration), a database, an
embedded processor,anembedded OS,andembedded software
with executable codes for the prognostics algorithms.
Probably the most straightforward implementation of such
embedded prognostic device with current technologies would
be to use some commercial embedded processor, running an
embedded RTOS with prognostic algorithms being implemen-
ted in programming languages like C/C++ and being executed
as concurrent tasks under the RTOS. However, considering that
prognostics algorithms are very distinct with regard to their
processing models, an optimized embedded system synthesis
can be achieved by exploring the design space evaluating
different customizable processor architectures. For instance, as
presented in (Beck & Carro, 2005) different usage of a
customizable DSP-based processor could lead to performance
improvements in the range of a factor of 4, with energy
reduction by a factor of even 10, at the price of extra area. So,
depending of the effective design goals (area, power dissipa-
tion, just performance or a combination of them all), the tuning
of the processor to the specific system it must work in can
provide a good balance within the design space, with a
reasonable goal. Moreover, since all the different micropro-
cessor descriptions are done in VHDL and synthesized of any
FPGA available in the market, dynamically changing them
whenever required by the upgrade of application allows for
easy adaptability without performance loss.
7. Concluding remarks
As discussed in the paper, looking from a top-down
perspective, modern manufacturing systems are challenged to
incorporate increasing capabilities of reconfigurability, self-x
and intelligence in order to be able to succeed in a very
competitive and global market, on which product variety and
complexity increase, product lifecycle shrinks, quality require-
ments increase, and profit margins decrease. Fortunately, when
considering a bottom-up perspective, one can identify that
considerable advances have been made in the last years in
advances are allowing the deployment of distributed and real-
time embedded computing architectures that can become key
enablers to the development of reconfigurable/intelligent
The paper described recent advances in DRES, and
presented some case studies that take benefit of some of these
new technological conditions. As technology evolves, the
immediate challenge is how to allow designers to deploy this
technology in the field. The test cases showed that not only this
is possible with available tools, but also that the research must
continue to cover new aspects that are being leveraged by
software and hardware technology advances.
This work has been partly supported by the Brazilian
research agencies CNPq, Fapergs, and FINEP.
Anke, J., & Neugebauer, M. (2006). Early data processing in smart item
environments using mobile services. In Proceedings of the IFAC 12th
Atherton, R. W. (1998). Moving Java to the factory. IEEE Spectrum, 1998, 18–
Balakrishnan, A., Kumara, S., & Sundaresan, S. (Jul 1999). Manufacturing in
the digital age: Exploiting information technologies for product realization.
Information Systems Frontiers, 1(1), 25–50.
Beck, A., & Carro, L. (2005). Dynamic reconfiguration with binary translation:
Breaking the ILP barrier with software compatibility. In Proceedings of the
42nd Design Automation Conference (pp. 732–737).
Beck, A. C., & Carro, L. (2004). A VLIW low power Java processor for
embedded applications. ACM SBCCI 2004 (pp. 157–162)1-58113-947-0.
Becker, L. B., & Pereira, C. E. (2002). SIMOO-RT—an object-oriented frame-
work for the development of realtime industrial automation systems. IEEE
Transactions of Robotics and Automation, 18(4), 421–430.
Bergamaschi, R. A., Bhattacharya, S., Wagner, R., Fellenz, C., Muhlada, M.,
White, F., et al. (2001). Automating the Design of SOCs Using Cores. In
IEEE Design & Test of Computers (p. 32–45).
Bernstein, P. A. (1996). Middleware: A model for distributed systems services.
Communications of the ACM, 39(2), 86–98.
Bollella, G., Gosling, J., & Benjamin, B. (2001). The Real-Time Specification
for Java, http://www.rtj.org/rtsj-V1.0.pdf.
Brinkschulte, U., Ungerer, T., & Becker, J. (2004). CARUSO: An approach
towards low power autonomic SoCs for embedded RT applications.
In Proceedings of the 18th IEEE Parallel and Distributed Processing
Brisolara, L., Becker, L. B., Carro, L., Wagner, F. R., & Pereira, C. E. (2005).
UML for SoC design. In Martin, G., & Muller, W. Eds. A comparison
between UML and function blocks for heterogeneous SoC design and ASIP
generation. Vol. 1 (pp.199–222).Springer. Chapter 9, ISBN 0-387-25744/6.
Burns, A., & Wellings, A. (1997). Real-time systems and programming lan-
Bussmann, S., Jennings, N. R., & Wooldridge, M. (2004). Multiagent systems
for manufacturing control: A design methodology. Springer-Verlag.
Christensen, J. H. (1994). Holonic manufacturing systems: Initial architecture
and standards directions. In Proceedings of the First European Conference
on Holonic Manufacturing Systems (pp. 1–20).
Approach. Heidelberg: Springer Verlag.
De Micheli, G., & Benini, L. (2002). Networks-on-Chip: A new paradigm for
systems-on-chip design. IEEE DATE’02—Design, Automation and Test in
Europe, Paris, 2002.
Densmore, D., Passerone, R., & Sangiovanni-Vincentelli, A. (September–
October, 2006). A platform-based taxonomy for ESL design. IEEE design
and test of computers, p. 359–374.
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 90
DFG (2004). DFG SPP 1183. Organic computing. http://www.organic-compu-
Draper, C. (1984). Flexible Manufacturing Systems Handbook, Automation and
Management Systems Division. The Charles Stark Draper Laboratory Inc
Filos, E. (2004). European research and policies for knowledge-driven innova-
tion. The example of industrial informatics. In Proceedings of the second
IEEE international conference on industrial informatics (INDIN’04) (pp.
Fletcher, M., & Brennan, R. W. (2001). Designing Holonic manufacturing
systems using the IEC 61499 (function block) architecture. IEICE Transac-
tions (Vol. E84-D, No. 10, p.1398–1401).
Gershenfeld, N. (1999). When things start to think. New York: Owl Books,
organizations: Strategies for enriching the customer. New York: Van
Go ¨tz, M., Rettberg, A., & Pereira, C. E. (2005). Towards run-time partitioning
of a real time operating system for reconfigurable systems on chip. In
Proceedings of IESS.
Hartenstein, R. (2001). A decade of reconfigurable computing: a visionary
retrospective. IEEE DATE (p. 642–649).
IEC. (2003). IEC 61784-1. Digital data communications for measurement and
control. Part 1. Profile sets for continuous and discrete manufacturing
relative to fieldbus use in industrial control systems.
IMS. (2006). NSF I/UCRC Center for Intelligent Maintenance Systems,
Ishii, K. (1998). Modularity: A key concept in product lifecycle engineering. In
A. Molina & A. Kusiak (Eds.), Handbook of life-cycle enterprise. Kluwer.
Ito, S. A., Carro, L., & Jacobi, R. P. (2001). Making Java work for micro-
controller applications. IEEE Design and Test of Computers, 18(5), 100–
Johnson, T., & Dausch, M. (2006). Sensor informatics for manufacturing. In
Proceedings of the IFAC 12th INCOM 2006.
Kephart, J. O., & Chess, D. M. (2003). The vision of autonomic computing.
Computer, 36(1), 41–50.
Keutzer, K., Malik, S., Newton, A. R., Rabaey, J., & Sangiovanni-Vincentelli,
A. (December 2000). System-level design: orthogonalization of concerns
and platform-based design. IEEE Transactions on Computer-Aided Design
of Integrated Circuits, 19(12), 1523–1543.
Kiritsis, D. (2004). Ubiquitous product lifecycle management using product
embedded information devices. In Proceedings of 2004 intelligent main-
tenance systems (IMS) international conference.
Koestler, A. (1989). The ghost in the machine. London: Arkana Books.
Koren, Y., & Ulsoy, A. G. (1997). Reconfigurable manufacturing systems.
Engineering research center for reconfigurable machining systems. Report
#1. University of Michigan, Ann Harbor.
Krapf, R., & Carro, L. (2003). Efficient Signal Processing in Embedded Java
Systems. ISCAS 2003 (p. IV-61–64). ISBN 0-7803-7762-1.
Lawton, G. (2002). Moving Java into mobile phones. Computer, 35(6),
Lee, J., Qiu, H., Ni, J., & Ad Djurdjanovic, D. (2004). Infotronics technologies
and predictive tools for next-generation maintenance systems. In Proceed-
ings of the 11th IFAC INCOM 2004.
Loyall, J. P., Schantz, R. E., Zinky, J. A., & Bakken, D. E. (1998). Specifying
and measuring quality of service in distributed object systems. In Proceed-
ings of IEEE ISORC ’98.
Luck, M., McBurney, P., & Preist, C. (2001). Agent technology: Enabling next
generation computing. A roadmap for agent-based computing, AgentLink.
Lyonnard, D., Yoo, S., Baghdadi, A., & Jerraya, A. (2001). Automatic gen-
eration of application-specific architectures for heterogeneous multi-pro-
cessor system-on-chip. DAC’01—design automation conference.
Mahalik, N. P. (Ed.). (July 2003). Fieldbus technology: industrial network
standards for real-time distributed control. Springer Verlag. ISBN:
Mar ˇı ´k, V., & Laz ˇansky ´, J. (2004). Industrial applications of agent technologies.
In Proceedings of the 11th IFAC INCOM 2004.
Mar ˇı ´k, V., & McFarlane, D. (2005). Industrial adoption of agent-based tech-
nologies.InIEEE intelligentsystems(Vol.20, No.1,p. 27–35).ISSN1094-
Mar ˇı ´k, V., & Pechoucek, (2001). Holons and agents: Recent development and
mutual impacts. In Proceedings of the 12th International Workshop on
Database and Expert Systems Applications.
McFarlane, D. (2002). Auto-ID based control—An overview. Whitepaper.
Mehrabi, M. G., Ulsoy, A. G., & Koren, Y. (2000). Reconfigurable manufactur-
ing systems: Key to future manufacturing. Journal of Intelligent Manu-
facturing, 11(4), 403–419.
Mentor. (2004). Seamless CVE. http://www.mentor.com/seamless.
Molina, A., Rodriguez, C., Ahuett, H., Corte ´s, J., Ramı ´rez, M., Jime ´nez, G.,
et al. (2005). Next-generation manufacturing systems: Key research issues
in developing and integrating reconfigurable and intelligent machines.
International Journal of Computer Integrated Manufacturing, 18(7),
Morel, G., Valckenaers, P., Faure, J. M., Pereira, C., & Diedrich, C. (2005).
Manufacturing plant control: Challengesand open issues. In Proceedings of
the 16th IFAC Triennial World Congress.
Neumann, P. (2007). Industrial communication: What is going on? Control
Engineering Practice (special issue with INCOM 2004 selected papers),
Nof, S. (2004) Collaborative e-Work and e-Mfg.: The state of the art and
challenges for production and logistics managers. Keynote paper at 11th
IFAC INCOM 2004. Salvador, Brazil. (In C. Pereira, G. Morel, & P.
Kopacek (Eds.). Information Control Problems in Manufacturing 2004.
Elsevier Science. ISBN-13: 978-0-08-044249-5).
O’Nils, M., & Jantsch, A. (2001). Device driver and DMA controller synthesis
embedded systems (Vol. 6, No. 2), Kluwer Academic Publisher.
Parunak, H. (1999). Industrial and practical applications of DAI. In G. Weiss
(Ed.), Multiagent systems: A modern approach to dist.artificial intelligence
(pp. 377–416). MIT Press.
Pedreiras, P., & Almeida, L. (2005). Approaches to enforce real-time behavior
in Ethernet.In R. Zurawski (Ed.), The industrial communication technology
handbook. Boca Raton, FL: CRC Press.
Pereira, C. E., & Mitidieri, C. (1999). Multi-agent systems from a real-time
perspective. Multi-agent-systems in production. ISBN 0-08-043657-9.
PROFIBUS Guideline. (2002). PROFInet architecture description and speci-
fication, Version 1.9. Karlsruhe: PNO.
Rammig, F., Gotz, M., Heimfarth, T., Janacik, P., & Oberthur, S. (2006). Real-
time operating systems for self-coordinating embedded systems. IEEE
Ravi, S., Raghunathan, A., Kocher, P., & Hattangady, S. (2004). Security in
embedded systems: Design challenges. Transactions on Embedded Com-
puting Systems, 3, 461–491.
Riedl, M., Diedrich, C., & Nauman, F. (2006). Event driven applications for
automation area. In Proceedings of the IFAC 12th INCOM 2006.
Rowson, J., & Sangiovanni-Vincentelli, A. (June 1997). Interface-based design.
In Proceedings of the DAC’97—Design Automation Conference.
Sangiovanni-Vincentelli, A., & Martin, G. (2001). Platform-based design and
Schantz, R., & Schmidt, D. (2001). Middleware for distributed systems:
Evolving the common structure for network-centric applications. Encyclo-
pedia of software engineering. Wiley&Sons.
Schantz, R., Loyall, J., Rodrigues, C., & Schmidt, D. (2006). Controlling
quality-of-service in distributed real-time and embedded systems via adap-
tive middleware. http://www.cs.wustl.edu/?schmidt/PDF/AM.pdf.
Schmek, H. (2005). Organic computing—A new vision for distributed
embedded systems. IEEE ISORC.
Schmidt, D. (2002). R&D advances in middleware for distributed, real-time,
and embedded systems. Communications of the ACM special issue on
Schmidt, D., Levine, D., & Mungee, S. (1998). The design and performance of
the TAO real-time object request broker. Computer Communications Spe-
cial Issue on Building Quality of Service into Distributed Systems 21(4).
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 91
SEEP. (2006). Sistemas Eletro ˆnicos Embarcados baseados em Plataformas.
Platform-based Embedded Systems Development. www.inf.ufrgs.br/?lse.
Shen, W., & Norrie, D. H. (1999). Agent-based systems for intelligent man-
ufacturing: A state-of-the-art survey. Knowledge and Information Systems,
Silva, E., Jr., Freitas, E., Wagner, F., Carvalho, F., & Pereira, C. E. (2006). Java
framework for distributed real-time embedded systems. In Proceedings of
9th IEEE International Symposium on Object and Component-Oriented
Real-Time Distributed Computing (ISORC) (pp. 85–92).
Stankovic, J. (1996). Strategic directions in real-time and embedded systems.
ACM Computing Surveys, 28(4), 751–763.
Synopsys. (2003). CoCentric System Studio. http://www.synopsys.com.
Van Brussel, H. (1994). Holonic manufacturing systems, the vision matching
the problem. In Proceedings of the First Conference on Holonic Manu-
Van Brussel, H., Wyns, J., Valckenaers, P., Bongaerts, L., & Peeters, P. (1998).
Reference architecture for holonic manufacturing systems: PROSA. Com-
puters and Industry, 37(3), 255–274.
Van Leeuwen, E. H., & Norrie, D. (1997). Intelligent manufacturing: Holons
and Holarchies. Manufacturing Engineering, 76(2), 86–88.
Vyatkin, V., Christensen, J. H., & Lastra, J. L. M. (2005). OONEIDA: An open,
object-oriented knowledge economy for intelligent industrial automation.
IEEE Transactions on Industrial Informatics, 1(1), 4–17.
Wehrmeister, M. A., Becker, L. B., & Pereira, C. E. (2004). Optimizing real-
time embedded systems development using a RTSJ-based API. Workshop
on Java Technologies for Real-Time and Embedded Systems—JTRES 2004,
Proceedings Springer LNCS (pp. 292–297).
Carlos Eduardo Pereira received the Dr.-Ing. degree in electrical engineering
from the University of Stuttgart, Germany in 1995, the M.Sc. degree in
computer science in 1990 and the B.S. degree in electrical engineering in
1987, both from the Federal University of Rio Grande do Sul (UFRGS) in
Brazil. He is an associate professor of the Electrical Engineering Department at
the Federal University of Rio Grande do Sul in Brazil, where he has served as
Dean of the EE Department from 1996 to 1998 and Coordinator for the
Graduate Research Program from 2002 to 2006. From 2000 to 2001 he was
a visiting researcher at the United Technologies Research Center (UTRC) in
Hartford, CT, USA, where he acted as Group Leader of the Embedded
Information Devices Group and has coordinated a group of 15 research
engineers involved with research projects for United Technologies companies,
such as Carrier, Otis, Pratt and Whitney, Sikorsky and UT Fuel Cells. Since
2005 he is acting as technical director for CETA—an Applied Research Center,
whose goal is to promote collaborative research work between academia and
industry, focusing on the areas of industrial automation, information and
communication technologies, and optimization of production processes. Prof.
Pereira’s research focuses on methodologies and tool support for the develop-
ment of distributed real-time embedded systems, with special emphasis on
industrial automation applications and the use of distributed objects over
computer-based systems. He is Chair of the IFAC Technical Committee on
Manufacturing Plant Control (TC 5.1). He is also an Associate Editor of the
Journals ‘‘Control Engineering Practice’’ – Elsevier and AtP International,
Oldenbourg. He has more than 150 technical publications on conferences and
journals and has acted as member of International Program Committees for
several conferences in the field of industrial automation, manufacturing,
industrial protocols, and real-time distributed object computing. He has been
the general chair of the 21st IFAC Workshop on Real-Time Programming,
WRTP’96, the 5th IFAC Workshop on Intelligent Manufacturing Systems,
IMS’98, the 2nd IFAC Workshop on Intelligent Assembly and Disassembly,
IAD’01 and the 11th IFAC Symposium on Information Control Problems in
Luigi Carro was born in Porto Alegre, Brazil, in 1962. He received the
electrical engineering and the M.Sc. degrees from Universidade Federal do
to 1991 he worked at ST-Microelectronics, Agrate, Italy, in the R&D group. In
1996 he received the Ph.D. degree in the area of computer science from
Universidade Federal do Rio Grande do Sul (UFRGS), Brazil. Prof. Carro is
presently at the Applied Informatics Department at the Informatics Institute of
graduate and undergraduate levels. He is also a member of the Graduation
Program in Computer Science at UFRGS, where he is responsible courses on
embedded systems, digital signal processing, and VLSI design. His primary
research interests include embedded systems design, digital signal processing,
mixed-signal and analog testing, and rapid system prototyping. He has pub-
lished more than 120 technical papers on those topics and is the author of the
books Digital systems Design and Prototyping (in Portuguese) and Fault-
Tolerance Techniques for SRAM-based FPGAs. He has served as Technical
Program Committee member of several conferences, like DATE, VTS, ETS,
IESS + CODES, FPL, SAMOS and RAW.
C.E. Pereira, L. Carro/Annual Reviews in Control 31 (2007) 81–92 92