See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228649965

Distributed Control Systems (DCS) Group: Activity Report

Article · June 2006

CITATION READS
1 4,044

10 authors, including:

Pau Martí Manel Velasco

Universitat Politècnica de Catalunya Universitat Politècnica de Catalunya
120 PUBLICATIONS   1,484 CITATIONS    89 PUBLICATIONS   906 CITATIONS   

SEE PROFILE SEE PROFILE

Camilo Lozoya José Yépez

Tecnológico de Monterrey Universitat Politècnica de Catalunya
36 PUBLICATIONS   182 CITATIONS    24 PUBLICATIONS   219 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

EVENTS - Event-Driven Embedded and Networked Control Systems View project

Automation for efficient use of water on agricultural irrigation systems View project

All content following this page was uploaded by José Yépez on 22 May 2014.

The user has requested enhancement of the downloaded file.

 Distributed Control Systems (DCS) Group: Activity Report

Pau Martı́, Manel Velasco, Josep Guàrdia, Rosa Castañé, Camilo Lozoya, Frederic Pérez,
José Yépez, Jordi Ayza, Ricard Villà and Josep.M Fuertes

Automatic Control Department, Technical University of Catalonia, Barcelona, Spain, e-mail: [email protected]

Abstract— This report presents an overview of the recent projects. Finally, section V concludes this report.
and on-going research and technology transfer activities of
the Distributed Control Systems (DCS) group. DCS research II. R ESEARCH ACTIVITIES
focuses on the study of the interaction of control systems and
real-time computing and communications systems for build-
The goal of the DCS research group is to develop so-
ing efficient embedded control systems, networked control
systems, and Quality-of-Service-aware systems. Research lutions for (networked) systems with strict real-time re-
advances are transfered as technology transfer projects to quirements (e.g., control systems). This usually involve
private companies mainly working in the automotive and au- the study of resource-constrained systems.
tomation sector. Mass market products are subject to hard economic
Index Terms— Control Systems, real-time systems, com- constraints and, also, many high-tech products are subject
munication systems to dimensional limitations or must work with restricted
communication bandwidth, processor capacity or battery
I. I NTRODUCTION power. This suggests that computing resources (e.g.,
processor, bandwidth, battery) have to be exploited effi-
This report presents an overview of the recent and on- ciently taking into account the application demands while
going research and technology transfer activities of the providing the best operation. In this context, feedback
Distributed Control Systems (DCS) group. DCS is a sub- mechanisms become fundamental. As a consequence,
group of the Grup de Recerca en Robòtica Intel.ligent control, computing, and communication have to be jointly
i Sistemes (GRINS) at the Universitat Politècnica de considered in any successful design.
Catalunya (UPC). DCS staff belongs to the Departament When considering integrated approaches, two main
d’Enginyeria de Sistemes, Automàtica i Informàtica In- (non independent) disciplines apply: feedback of comput-
dustrial (ESAII) of UPC. DCS webpage can be found at ing systems and computing of feedback systems. The for-
http://dcs.upc.edu. mer refers to the application of control theory to comput-
DCS research focuses on the study of the interaction of ing systems in order to improve their performance. The
control systems and real-time computing and communica- later refers to the application of novel (real-time) comput-
tions systems for building efficient embedded control sys- ing techniques to improve control systems performance.
tems, networked control systems, and Quality-of-Service- In addition, this may require new analysis and design tech-
aware (QoS aware) systems. Our research activities are niques for control systems.
mainly financed by public funds like Spanish research Our research activities tackle several aspects of both
projects, e.g., projects from the Comisión Interministerial disciplines:
de Ciencia y Tecnologia (CICYT, Interministerial Com- • control theory
mission on Science and Technology) of the Spanish gov- • real-time computing of distributed feedback control
ernment, and through the participation in european net- systems
works of excellence. • feedback of computing systems
Research advances are transfered by technology ap- • analysis and design tools
plied projects to private companies mainly working in the Some of the work described next has been carried out
automotive (including car and train industry) and automa- together with (listed in alphabetical order) Prof. Luis
tion sector. Future projects may cover the consumer elec- Almeida (University of Aveiro, Portugal), Prof. Guillem
tronics and other sectors that require QoS aware comput- Bernat (York University, UK), Prof. Scott Brandt (Univer-
ing. sity of California at Santa Cruz, USA), Prof. Giorgio But-
The rest of this report is organized as follows. Section tazzo (Scuola Superiore S. Anna, Pisa, Italy), Prof. Anton
II presents our research activities. Section III describes Cervin (Lund University, Sweden), Prof. Gerhard Fohler
the funding projects that sustain our research. Section IV (University of Kaiserslauten, Germany), Prof. Daniel
presents recent, on-going and future technology transfer Mossé (University of Pittsburgh, USA), and Prof. Krithi
Ramamritham (Indian Institute of Technology Bombay, bly, develop new ones. A first result has been to obtain an
India). interpretation, inexistent to the authors knowledge, of the
All the research activities listed next are on-going re- added pole when discrete time system with time delays
search projects or new starting projects. Some results have are considered.
been already published, or they are being considered for An step further will be considering the use of discrete-
publication in journals or conferences, others have been time event systems to approach existing problems.
described in technical reports, and the left overs have not
A.3 Treating jitters as perturbation
been reported yet.
Jitters due to networking and computing introduce
A. Control theory degradation in control loops. We plan to investigate
whether they can be included in the control loop as pertur-
The research goal of the activities here described is to
bations, and design controllers capable of rejecting such
adapt existing control theoretical results and to develop
perturbations.
new control theory giving answers to the demands of
novel control application architectures such as multitask- A.4 Collaborative control
ing real-time control systems or networked control sys-
Networked control systems lack a central entity ca-
tems.
pable of coordinating the control operations of the dis-
A.1 Stability tests for systems with varying sampling tributed nodes. This demands distributed coordination of
rates and time delays those nodes performing control activities, which can be
approached using collaborative control techniques.
In an Networked Control Systems or Real-Time Con-
trol Systems, varying sampling rates and time delays may A preliminary result has been to develop a state space
bring control loops to instability. This situation appears formulation for networked control loops using non-linear
because, at the controller design stage, the specific tim- control techniques that permits control nodes to self-
ing of the networked or computing implementation is not assign bandwidth in a coordinated manner [3].
usually considered. For example, controllers are designed
B. Real-time computing of distributed feedback control
assuming a constant sampling period and time delay. Un-
systems
fortunately, at run-time, sampling instants may not be
equidistant and time delays may vary. The main objective of the activities described next is to
An approach to system stability develop and exploit novel contributions of real-time sys-
• is based on modeling the irregular sampling and vary- tems theory in order improve control applications perfor-
ing time delay as random variables with known ex- mance. A key aspect in these works is that computing
pectation. With this model, the evolution of the sys- and communication resources are limited, that is, control
tem can be seen as a sequence of random state vectors applications belong to the class of resource constrained
generated by the system closed-loop matrix. Then, systems.
we can derive stability conditions in terms of conver-
B.1 Jitter compensation and flexible timing constraints
gence of a sequence of random variables [1].
• is based on modeling the variability using interval Traditionally, real-time control systems have been de-
algebra methods. Then, by applying robust control signed in two stages: first, control design, and then, its
techniques to a wrapping model we can obtain a suf- computer implementation. For resource-constrained real-
ficient stability condition that allows us to test the time embedded systems with multiple control loops, this
stability of a (networked) closed-loop system subject often leads to sub-optimal solutions in terms of both sys-
to bounded varying sampling rates and time delays. tem schedulability and controlled system’s performance.
We can show that just checking the stability of one Motivated by this, we have developed a flexible analysis
time-invariant matrix is enough to guarantee stability and design framework for integrated scheduling and con-
of the original system [2]. trol in real-time systems.
Further work will focus on Lie algebra methods applied to The new analysis and design approach to discrete-time
the stability of these systems. controllers, named compensation approach, goes beyond
the classic discrete-time control theory timing assump-
A.2 Mathematical models for sampled data systems tions of equidistant sampling and equidistant actuation.
Existing standard mathematical models for discrete The controllers resulting from our compensation approach
time control systems fail at providing the best control per- support closed-loop implementations with irregular sam-
formance when they are used to design control laws for pling and varying time delays. The new timing assump-
real-time tasks. We will study existing models, and possi- tions behind the compensation approach motivate the need
 Schedule 400
repeated

..... 350 Static periods

T1

T2 .....
300
T3 .....

250 FBS assuming stationary noise

0 50 100 150 200 250 300 350 400

Schedule 200
repeated

T1 ..... 150 Estimator 1

T2 .....
Estimator 2
.....
100
T3

50 Known noise variances

0 50 100 150 200 250 300 350 400 (ideal case)

0
0 5 10 15 20 25 30 35 40 45
Fig. 1. Scheduling the same task set with (top) fixed timing con- Time (sec)

straints (giving an unfeasible schedule) and (bottom) flexible

timing constraints (giving a feasible schedule) Fig. 2. Cumulative errors for five different simulations. Both
proposed estimators come close to the performance obtained
when the true noise variances are used by the feedback
for flexible timing constraints for control tasks, beyond scheduler.
task periods and deadlines. With the flexibility given by
the compensation approach, we eliminate the control per-
formance degradation caused by scheduling induced jit- tion of networks within a control loop. Using this model
ters. In addition, flexible timing constraints improve task control signals are computed using old plant states com-
set schedulability (Figure 1 shows a simple example) and pared to the time they are applied to the plant, which may
offer the possibility of implementing flexible scheduling not provide the best “control”.
or resource allocation policies that are shown to improve To overcome this problem, we will study novel con-
overall control performance [5]. trol models to improve control performance compared to
B.2 Optimal sampling period for real-time control tasks existing standard control models for real-time control sys-
tems that can be found in the literature. Specifically, we
In embedded systems, the computing resources are of-
will focus the input/output synchronization properties of
ten scarce, and many controllers may have to share the
control models and their impact on the computation of
same computer. Let us assume that n independent control
control signals. From this study, we expect to offer new
tasks should be scheduled on one CPU. We will study the
control models and the corresponding algorithms aimed at
problem of optimal period assignment, where the goal is
improving control performance.
to assign sampling periods to the controllers so that the
overall control performance is maximized, subject to a A preliminary result [7] using a prediction method is
constraint on the CPU utilization. shown in Figure 3.
The main objective is to develop a feedback scheduler
that periodically assigns new sampling periods to the con- 0.35

trollers based on estimates of the plant states and the in- o Modelling the
0.3 time delay
tensities of the noise processes, while optimizing control
performance. 0.25
+ Predictor Model

Preliminary results shown in Figure 2 show the cumu-

lative error of three controllers when scheduled using dif- 0.2
y

ferent techniques. The two estimator curves implement 0.15

the latest feedback policies that we have developed.

New results on this activity have been submitted to [3] 0.1

and [4].
0.05

B.3 Control models for real-time control tasks

0
0 50 100 150 200 250 300
A realistic control model found in control text-books time(s)

includes a time delay. This delay is used to model the

Fig. 3. Predictor model performance
input/output latency that appears for example due to the
computation of the control algorithm or due to the inser-
 new control law requires too much time to be performed at
runtime, a set of controllers has to be designed off line for
message on bus from pendulum 2
different rates and the system has to switch to the proper
controller in the presence of an overload condition.
We have presented a method for reducing the number
denied access
of controllers to be designed off line, while still guaran-
message on bus from pendulum 1 teeing a given control performance
. The proposed approach has been integrated with the elas-
tic scheduling theory to promptly react to overload con-
denied access ditions. The effectiveness of the proposed approach has
been verified through extensive simulation experiments
performed on an inverted pendulum. In addition, the
0 50 100 150 200 250 300 350 400 450 500
Messages on bus method has been implemented on a real inverted pendu-
lum.
h h h 1.6
Envelope
Pendulum 1 messages Envelope error
1.4
Pendulum 2 messages PRFs
1.2 QoC min
Fig. 4. Bus schedule 1

QoC
0.8

B.4 Networked Control Systems application profile 0.6

0.4
In recent years, approaches to control performance
optimization for resource-constrained systems (e.g., net- 0.2
Tmin Tmax
worked embedded systems) have been receiving increased 0 0.1 0.2 0.3 0.4 0.5 0.6
attention. Most of them focus on theory, whereas practical Periods(s)
aspects are omitted. We will work on a Controller Area
Network (CAN) application profile that permits imple- Fig. 5. Baseline graphical algorithm
menting networked embedded systems with control per-
formance optimization quickly. First, we should justify Figure 5 illustrates graphically the main algorithm that
the selection of the network, CAN. Afterwords, the spe- we have developed. It is based on studying the perfor-
cific application profile will be carried out. We finally mance of controllers at different sampling rates, and then
shall show simulation results validating the applicability derive the minimum set of controllers.
of the new profile.
C. Feedback of computing systems
Preliminary simulation results [8] shown in Figure 4
show asynchronous message transmissions based on state The application of feedback control techniques to en-
feedback information of two network-controlled inverted hance computing systems performance has been proved
pendulums. to be a promising approach. Companies such Google or
Amazon already use these techniques in their web-server
B.5 Feedback scheduling of tracking systems
systems.
So far, all the existing research on feedback scheduling
has focused on regulation problems. We plan to extend C.1 QoS aware video streaming for consumer electronics
existing results to tracking systems. In-home multimedia entertainment platform makers
face the problem of channel saturation. This happens with
B.6 Overload management in real-time control systems multiple video transmissions, and severely limits the qual-
Transient overload conditions may cause unpredictable ity of service. The problem is aggravated by the tendency
performance degradations in computer controlled sys- to use more receptors, many of them portable. An effec-
tems, if not properly handled. To prevent such problems, tive method will be developed to manage resources in an
a common technique adopted in periodic task systems is adaptive video transmitting environment. It must assure a
to reduce the workload by enlarging activation periods. In quality of service to the receptor notwithstanding the vary-
a digital controller, however, the variation applied on the ing bandwidth and user’s preferences.
task period also affects the control law, which needs to The key in achieving this will be the use of resource
be recomputed for the new activation rate. If computing a managing techniques based in Control Theory and Op-
timal Scheduling, often known as feedback of comput- function of the behavior of the controlled plants. In or-
ing systems. A global resource manager, using an esti- der to properly analyze how tasks periods are changed, we
mation/prediction of the available bandwidth and the user have added an new output port, “H”, to the kernel block
preferences, will assign a bandwidth to each emitter. Each (Figure 7). To facilitate the implementation of these ap-
emitter will process the video to be transmitted so as to fit proaches, we have modified the kernel and added a func-
it into its assigned bandwidth and assure its quality. tion “ttAttachPreScheduler” that permits to call a generic
Some preliminary results on bandwidth estimation us- user-defined function that we call pre-scheduler(e.g., an
ing Kalman filtering techniques can be seen in Figure 6 optimization procedure) that can re-assign tasks periods
before each scheduler invocation. Minor modifications
Integral Absolute Error have been also included [9].
0.35
Mean
Median
Max
Kalman

0.3

0.25
Fig. 7. Kernel block with the new “H” port

D.2 Enabling Feedback Scheduling in SHaRK

0.2
5 10 15 20 25 As in TrueTime, SHaRK (Soft Hard Real-Time Kernel,
http://shark.sssup.it/) has been similarly modified to allow
Fig. 6. Kalman prediction
easy implementation of feedback scheduling policies.
D.3 CICLIC: A software tool to Generate Feasible Cyclic
C.2 Dynamic Voltage Scaling: a feedback approach
Schedules
Dynamic Voltage Scaling (DVS) is a set of techniques A common way of implementing periodic real-time
that allow to specify the processor velocity in such a way systems is by means of a cyclic schedule. This scheduling
that energy can be saved. We plan to apply feedback tech- method is characterized by having a high degree of de-
niques when DVS is performed at the processor level as terminism, a high degree of predictability, a high degree
well as at the memory level (main memory and cache of reliability, and implementation simplicity. Also, it is a
memory). very well-known method broadly used in industrial envi-
ronments.
D. Analysis and design tools
The main problem this type of schedules may have is
Most of the reserach described above requires specific the lack of flexibility, since any change in the group of
analysis and design tools, as well as specific executing tasks or in its temporal constraints forces to repeat the de-
platforms (e.g., real-time kernels). Next, we describe sign of the execution plan. In solving these problems, it
some of our contributions in this field. is necessary, and most of the time, essential, to have tools
that may help system designers to create a cyclic sched-
D.1 Enabling Feedback Scheduling in TrueTime
ule automatically. We have developed a tool having this
TrueTime is a MATLAB/Simulink based tool function that we named CICLIC [10]. The design mecha-
(www.control.lth.se/∼dan/truetime/) that permits simula- nism is based on algorithms of exhaustive search that use
tion of the temporal behavior of a multitasking real-time heuristic rules in optimizing searching paths.
kernel executing controller tasks. The tasks are processes Figure 8 shows an screen shoot of the CICLIC tool.
that control continuous-time plants modeled as ordinary
Simulink blocks. We have performed simple modifica- D.4 Real-time kernel for distributed control applications
tions of the TrueTime simulator aimed at facilitating the One of the main problems in the analysis and design
implementation of feedback scheduling policies. of distributed control applications is the lack of hardware
Feedback scheduling refers to those flexible and adap- platforms integrating real-time processing and real-time
tive scheduling or resource allocation techniques for real- communication, as well as, interfaces for plant control.
time control systems that specify tasks executions as a We have recently ported an existing real-time kernel
 B. European network of excellence
Our research activities at the european level take place
within the FP6 IST-004527 ARTIST2 Network of Excel-
lence on Embedded Systems Design (http://www.artist-
embedded.org/FP6/), starting September 2004 with a du-
ration of 48 months. The objective of ARTIST2 is to
strengthen European research in Embedded Systems De-
sign, and promote the emergence of this new multi-
disciplinary area. We actively participate in the Adaptive
Real-Time cluster. Adaptive real-time is a recent approach
to embedded systems design, where temporal constraints
Fig. 8. Screen view
can be relaxed, which allows optimized use of resources.
This includes applications where managing the Quality
of Service (QoS) is essential, such as telecommunication
(developed by the research group of Prof. Luis Almeida, systems, multi-media, and wide-area networked applica-
http://www.ieeta.pt/lse/) build on top of a PIC18F258 tions. In this relatively new area, there is a recognized
microprocessor to a PIC18F458 microprocessor. The lack of design theory and tools.
later incorporates an integrated Controller Area Network ARTIST2 is the continuation of the FP5 IST-2001-
(CAN) communication controller, which will allow to 34820 Network of Excellence on Advanced Real-Time
jointly study tasks and message scheduling. In addition, Systems (ARTIST, http://www.artist-embedded.org/).
these type of microprocessors can be easily interfaced to ARTIST main objectives were to coordinate the RD
plants, permitting the study of distributed control applica- effort in the area of Advanced Real-time Systems so
tions. as to a) improve awareness of academics and industry
in the area, especially about existing innovative results
and technologies, standards and regulations, and b)
III. F UNDING PROJECTS
define innovative and relevant work directions, identify
obstacles to scientific and technological progress and
A. Spanish CICYT propose adequate strategies for circumventing them. The
main output of ARTIST is the The ARTIST Roadmap for
Our research group has been financed by several CI- Research and Development [11].
CYT projects. The last project was DPI2002-01621
CRITICO: Integrated Co-design of Controllers and Real- IV. T ECHNOLOGY T RANSFER
Time Schedulers. Application to Distributed Control Sys-
tems The main objective of this project was the study and In the following, we describe the technology applied
the experimentation of the hard real-time characteristics projects that outcome from our research activities.
in distributed control systems and the integrated design of
the control algorithm together with the real-time sched- A. Automotive sector
uler.
Whenever the reliability and the quality of the control
We have submitted a proposal for a new CICYT remain guaranteed, the use of processor networks in the
project: Video Streaming with adaptive and mobile re- automobile makes possible to achieve sophisticated con-
sources (STREAM VIDEO). This project is motivated by trol tasks. Also, the automotive industry is very compet-
the saturation of the communications in In-home multi- itive and it bears much in mind the cost of all the com-
media entertainment, that consumer electronics makers ponents that it uses. In the processor network, the cost
face. This happens with multiple video transmissions, and depends on the optimal use of the resources: processor
severely limits the quality of service. The project’s ulti- memory, processor execution time and network occupa-
mate aim is the use of feedback computing techniques to tion. For this reason, one of the aims in this sector is to
ensure a guaranteed level of service delivery across com- look for the maximum efficiency in the use of resources
plex inter-operable environments, as well as an optimized guaranteeing the reliability and the quality of the control.
use of underlying delivery network bandwidth/QoS char- A applied project, Analysis of Processor Efficiency,
acteristics, allowing for availability of high added value Scheduling Optimization and Improvements in the Utiliza-
scalable multimedia contents and programs, and seamless tion of Control Resources with Lear Automotive focused
device connectivity. on an automotive control box, executing a defined task
set, and adopting OSEK specifications for real-time oper- C. Automation sector
ation. The main goal has been to optimize processor time
The increasing use of flexible and powerful distrib-
use in the car control boxes (Figure 9) so as to be able to
uted (control) systems and the expanding role of field-
include new algorithms without additional cost and main-
bus systems and Internet as communication backbones are
taining the required control quality. The control quality
presently characterizing the industrial automation field.
This tendency is still presenting some drawbacks, one of
them still being the integration of different standards and
technologies. The lack of easy integration jeopardizes the
operation of distributed automation that requires soft and
hard real-time constraints.
Working toward finding solutions to these kind of prob-
lems, our group has been working in several projects.
The latest has been a collaboration with Aplielec, named
Advanced communications systems for a SCADA envi-
Fig. 9. Project scenario ronment. Aplielec is a company devoted to automation
projects. In particular, they have developed a product for
was assured by accomplishing the specific time require- supervision and process control, which integrates several
ments. We considered different task scheduling methods, components like SQL Server, OPC Server, Modbus/TCP
including dynamic scheduling and, finally, we applied the based PLC’s, and Industrial Ethernet (see Figure 10 for an
best scheduling method compatible with OSEK specifica- scheme of the architecture). The operation between these
tions. components rises integration problems, implying ineffi-
We have established a set of rules to apply so as to as- cient communications in specific application scenarios.
sign enough execution time to each task and thus guaran-
tee the meeting of the required specifications.

B. Railway sector

Reliability and security in trains is a must in actual

society. Present deployment of train structure can dis-
tinguish between high traffic lines and low traffic lines.
From those later, usually the train lines are with a single
track, shared by trains in both directions, and local deriva-
tions that allow crossing trains. Controlling the tracks’ Fig. 10. Project scenario
use is actually done mainly by human, using visual con-
trol and telephonic communication. Even that, accidents Our contribution has been to identify bottlenecks in the
still ocurr. exiting architecture, and to develop a novel communica-
To face this kind of problems, there is a project, the tion architecture that has overcome some of the mentioned
BLOCKSAT project, to design a system with on-board problems, while maintaining the original functionalities.
equipment, that collects continuously information about
train situation and movement, and a central subsystem that V. C ONCLUSIONS
tracks in real-time the current situation of all the trains in
This report has presented the main on-going activities
a line and potentially would help to obtain a more secure
of the Distributed Control Systems group.
transportation. The project is liedered by Sener, a com-
pany with experience in different transportation sectors,
R EFERENCES
and conducted through K2M, an initiative from the UPC
to develop advanced projects. In the project there is in- [1] M. Velasco, P. Martı́, R. Villà, J.M. Fuertes and M. Monroig, ”A
volvement from the dept. of TSC, in the part of GSM and Probabilistic Approach to the Stability Analysis of Real-Time Con-
trol Systems,”, In proceedings of the 16th IFAC World Congress,
GPS communications, from the dept. of EE for the de- Praha, Czech Republic, July, 2005.
velopment of the hardware platform, and from the dept. [2] M. Velasco, P. Martı́, R. Villà and J.M. Fuertes, ”Stability of Net-
of ESAII, for the Real Time software development, the worked Control Systems with Bounded Sampling Rates and Time
Delays,”, In proceedings of the 31th Annual Conference of the IEEE
hardware-software integration, and for the Validation and Industrial Electronics Society (IECON05), Raleigh, USA, Novem-
Verification parts of the project. ber, 2005.
 [3] P. Martı́, C. Lin, S. Brandt, M. Velasco and J.M. Fuertes, ”State session of the 18th Euromicro Conference on Real-Time Systems.
Feedback Based Resource Allocation for Resource-Constrained [8] M.Velasco, P.Martı́, R.Castañé, J.Guàrdia, and J.M. Fuertes, ”A
Control Tasks,” submitted to the Real-Time Systems journal. CAN Application Profile for Control Optimization in Networked
[4] A. Cervin, D. Henriksson, R. Castañé, P. Martı́ and M. Velasco, Embedded Systems,” submitted to the 32nd Annual Conference of
”Optimal On-Line Sampling Period Assignment Assuming Non- the IEEE Industrial Electronics Society.
Stationary Noise Processes,” submitted to the 45th IEEE Conference [9] J.Guàrdia, P.Martı́, M. Velasco and R. Castañé, ”Enabling Feed-
on Decision and Control back Scheduling in TrueTime.,” Research report ESAII-RR-06-05,
[5] P. Martı́, M. Velasco, J.M. Fuertes, G. Fohler and K. Ramam- March 2006
ritham, ”Integrated Scheduling and Control in Real-Time Systems: [10] J. Yépez, J. Guàrdia, M. Velasco, J. Ayza, R. Castañé, P. Martı́ and
An Analysis and Design Framework,” submitted to IEEE Transac- J.M. Fuertes, ”CICLIC: A Tool to Generate Feasible Cyclic Sched-
tions on Computers ules,”, in proceedings of the 6th IEEE International Workshop on
[6] G. Buttazzo, M. Velasco, P. Martı́, ”Quality-of-Control Manage- Factory Communication Systems.
ment in Overloaded Real-Time Systems,” submitted to IEEE Trans- [11] J.M.Fuertes, P.Martı́, et al. ”Adaptive Real-time Systems,” In Em-
action on Computers bedded Systems Design - The ARTIST Roadmap for Research and
[7] R. Castañé, P. Martı́ and Manel Velasco, ”A Novel Control Model Development. Series: Lecture Notes in Computer Science, Vol.
for Real-Time Control Tasks,” submitted to the work-in-progress 3436, 2005, Springer. ISBN: 3-540-25107-3.

View publication stats