Open access peer-reviewed chapter

Issues on Communication Network Control System Based Upon Scheduling Strategy Using Numerical Simulations

By Oscar Esquivel-Flores, Héctor Benítez-Pérez and Jorge Ortega-Arjona

Submitted: December 6th 2011Reviewed: May 11th 2012Published: September 19th 2012

DOI: 10.5772/48578

Downloaded: 1770

1. Introduction

Nowadays, industry has successfully used Network Control Systems (NCS) therefore several lines of research have arisen. A NCS is a current application of a Real-Time Distributed Systems (RTDS), composed of a number of nodes capable of developing a complete control process. In these systems several nodes exchange information through a communication network to achieve specific control goals, nevertheless network traffic increases. This affects the overall system performance. Several approaches have been developed to satisfy requirements of both control and communication performance. Particularly, some methodologies focus on saving bandwidth, one of such methodologies is network scheduling. The objective of this methodology is the accurately use of the computing resources. NCS research is categorized into two main parts [5]:

  1. Control of network: Study and research on communications and networks to make them suitable for real-time NCS, e.g. routing control, congestion reduction, efficient data communication, networking protocol.

  2. Control over network: This area deals with control strategies and control systems design over the network trying to minimize the effect of adverse network parameters on NCS performance such as network delay.

These systems have many challenges to maintain the the Quality of Service (QoS) and Quality of Control (QoC). In the networks, QoS is the idea that transmission rates, error rates, and other characteristics can be measured and improved. The QoS can be degraded due to congestion and interference.

1.1. Architecture and overview of a NCS

There are two general NCS configurations, Direct structure and Hierarchical structure [14]. Direct structure consists of a controller and a remote system containing a physical plant, sensors and actuators. The controller and the plant are physically located at different points and are directly linked by a data network in order to perform remote closed-loop control. Fig. 1 presents a schematic diagram of a direct structure of a NCS [15].

Figure 1.

Schematic diagram of a direct structure of a NCS

1.2. Effect of sampling period on NCS

Network scheduling deals with to elect a sampling rate, aiming to reduce the number of data transmitted over the network. The effectiveness of the control system depends on such a sampling rate [8, 910]. A region is acceptable in networked control performance terms if it is contained within two sampling rate boundaries, which can be statistically determined.

The use of a common-bus network architecture and a particular network protocol introduces different forms of time delay uncertainties between, sensors, actuators and controllers [8]. Hence, it is quite important to explore different network protocols and network scheduling strategies, before to implement the RTDS, in order to obtain a desired control performance.

For networked control, the minimum transmission frequency (fm)is necessary to guarantee good system performance without decreasing the network performance. As the transmission frequency increases the system performance improves; however, the load on the network also increases until a maximum transmission frequency (fh)is reached, then the system performance decreases because the network performance is overloaded.

This phenomenon represents non-linear situations with respect to sudden changes in state of the network, failure situations, or saturation in the channel or traffic, among others. However; it is possible to propose a linear model in the context of proper use of the network, thereby deferring the modeling of nonlinearity in these systems until future work.

The main objective of this work is to explore several issues on communication network scheduling of a RTDS, besides to implement a particular network scheduling strategy to evaluate its effectiveness using numerical simulations. This is presented using a simulated case study, based upon a 2-DOF helicopter simulation benchmark [13]. This simulation provides an approximation to system response, in which, for demonstration purposes, the main results are obtained for a typical fault scenario. Thus, for this simulation, a scheduling strategies is implemented using TrueTime [1, 2] performing dynamic scheduling. Several researches have focused on control over network, shared-network control systems have special interest.

2. Real-time simulation tool TrueTime

This section gives a brief overview of TrueTime simulator and exposes basic examples to initialize typical TrueTime blocks.

According to Cervin et al., nowadays simple embedded control systems often contain a multi-tasking real-time kernel and support networking, besides time control algorithm and control software designs need to be considered together. Thus new computer-based tools for real time and control design are needed [1]. Networked control loops consist of sensor, actuator and control calculations residing on different nodes; within the individual nodes the controllers are implemented as one or several tasks on microprocessor with a real-time operating system, this operating system typically uses multiprogramming to execute various tasks. Communication bandwidth and CPU time can be considered as shared resources for which the task compete. Different sources of temporal nondeterminism as execution times of the tasks or communications delays affect the control performance, nevertheless this nondeterminism can be reduce by the accurate choice of implementation platforms. The constraints of the implementation platform must be considered in systems with limited computer resources [1], therefore some tools are available to analyze and simulate the effects of temporal constrains affects control performance.

TrueTime [1, 2, 6, 7, 12] is a simulator for networked and embedded control system based on Matlab/Simulink, it has been developed at Lund University since 1999 [2]. TrueTime can be used as an experimental platform for research on dynamic real-time control systems. For instance, it is possible to study compensation schemes that adjust the control algorithm based on measurements of actual timing variations [1]. TrueTime make it possible to study more general and detailed timing models of computer-controlled systems. TrueTime can be used: to investigate how timing nondeterminism affects the system behavior, to develop new outlines to adjust control parameters dynamically, to experiment new approaches as codesign of control and network scheduling and to simulate control systems based on event-driven task [1].

The simulator software consists of a Simulink block library, the kernel block simulates a Real-Time kernel executing user-defined task and interrupt handlers. To communicate kernel blocks (nodes) several network blocks may be used, thus it makes quite simple to develop networked control system simulations.

2.1. The kernel block

A computer node is simulated using TrueTime kernel block, this node has a generic real-time kernel, A/D and D/A converters, and network interfaces. An initialization scrip is used to configure the block, in this script it is possible to create several objects as task, timers, interrupt handlers, semaphores, etc., this objects establish the software executing in the computer node. The kernel continuously calls the code functions of the tasks and interrupt handlers. Either Matlab m-files code or C++ language may be used to write initialization scripts and the code functions, the main advantage to use C++ is the speed, nevertheless m-file code is very easy to use. Several scheduling policy is able to use in TreTime kernel block, these can be fixed-priority scheduling and earliest-deadline-first scheduling and custom scheduling policies [2].

The task is the main construction in the TrueTime environment, this object is used to simulate periodic and aperiodic activities, for example controller and I/O tasks can be periodic and communication and event-driven controller can be aperiodic tasks. A set of attributes and a code function define a task; attributes as name, release time, worst-case execution time, budget, relative and absolute deadlines, priority (if fixed priority scheduling is used), period (if the task is periodic). Release time and absolute deadline are attributes constantly updated bye the kernel during simulation, while period and priority are kept constant, although can be changed by callas to kernel primitives trough execution [1]. An example of the definition of a task is shown below:

The kernel primitive ttInitKernel() initializes a sensor node. The kernel is initialized by specifying the number of A/D and D/A channels and scheduling policy. The built-in priority function prioFP specifies fixed-priority scheduling. Rate monotonic prioRM, earliest deadline first prioEDF, and deadline monotonic prioDM scheduling are additional predefined scheduling policies [6].

Interrupts can be generated in two ways: An external interrupt is associated with one of the external interrupt channels of the computer block; when the signal of the corresponding channel changes value the interrupt triggers. The usefulness of this type of interrupt lies to simulate distributed controllers that execute when measurements arrive on the network. Internal interrupts work to construct timers, when a timer expires the interrupt is triggered. A user-defined interrupt handler is scheduled when an external or internal interrupt occurs. An interrupt, as a task, handles works but it is scheduled on a higher priority level. An interrupt handler is defined by name, a priority and a code function [?]. An example of a definition of a interrupt handler is as follows:

Cervin et al. [1, 2] mentions that simulated execution occurs at three distinct priority levels: the interrupt (highest priority), kernel and task (lower priority) levels. The execution may be preemptive or non-preemptive. At interrupt level, interrupt handlers are scheduled according to fixed priorities. At task level, dynamic-priority scheduling may be used. At each scheduling point, the priority of task is given by user-defined priority function which is a function of the task attributes, this makes it easy to simulate different scheduling policies. Predefined priority functions exist for most of the commonly used scheduling schemes.

The code associated with task and interrupt handlers is scheduled and executed bye the kernel while simulation progresses. The code may be divided in segments, which can interact with other tasks and with the environment at the beginning of each code segment. The simulated execution time of each segment is returned by the code function and can be modeled as constant, random, or even data-dependent [1]. During the simulation the kernel saves the current segment and calls the code functions with proper arguments. Execution resumes in the next segment when the task has been running for the time associated with previous segment [7]. An example of a sensor code is given bellow:

This function implements a simple sensor node. In the first segment, the plant is sampled using a execution time of.5 ms. In the second segment, the control signal is sent to the controller node. The third segment indicates the end of execution by returning a negative execution time. The structure data represent the local memory and is used to store the measured variable between calls to the different segments. The kernel primitives ttAnalogIn and ttAnalogOut can be perform A/D and D/A conversion. Besides A/D and D/A conversion, a large set of kernel primitives exist which can be called from code function [12].

Monitors and events support sincronization between tasks. Monitors are used to guarnatee mutual exclusion when accessing comon data. Events are associated with monitors to represent condition variables [1].

Different output graphs are generated by truetime blocks. Each computer block will produce two graphs: A computer graph will display the execution trace of each task and interrupt handler during the simulation. If the signal is high, it means that the task is running. A medium signal indicates that the task is ready but not running, a low signal means that the task is idle. Otherwise a monitor graph shows which tasks are holding and waiting on the different monitors during simulation [1].

2.2. The network block

The TrueTime network block simulates the physical layer and the medium-access layer of several local-area networks. CSMA/CD (Ethernet), CSMA/AMP (CAN), Round Robin (Token Bus), FDMA, TDMA (TTP), Switched Ethernet, WLAN (802.11b), and ZigBee (802.15.4) are some types of network supported [2]. The network blocks are mainly configured using blocks dialogs. Some parameters common to all types of networks are bit rate, the minimum frame size, and the network interface delay. For each type of network there are some parameters that specifie the number of nodes, data rate (bits/s), preprocessing delay, minimum frame size, maximum frame size, frame overhead, etc. The network blocks may be used having one kernel block for each node in the network. The tasks into the kernels can send and receive arbitrary Matlab structure arrays over the network using ttSendMsg and ttGetMsg kernel primitives. This way is to quite flexible but requires to program some routines to configure the system. An useful network scheduler viewer shows the network activity for all nodes involved. An overview of all Truetime’s primitives can be found on [12].

3. Frequency transmission scheduling

In this section a formal definition of task model is provided and it gives an overview of the control of frequency transmission in a distributed systems and how it impacts over quality performance.

3.1. Task model

Chen et al. [3] gives a formal definition of system model of a typical distributed systems consisting of a set of processors and a set of tasks. A distributed system are characterized as follows. A set of processors χ=S1,S2,...,Smwhere χis the processor set, Siis the i-thprocessor and mis the total number of processors. In this model, all processors are assumed to be identical to assure same execution time for each task on different processors. It is also supposed that enough processors are provided. A set of primary copy of real-time tasks τ=Ϣ1,Ϣ2,...,Ϣn, where Ϣi=ci,pi, i=1,2,...,n. Here τis the set of tasks, Ϣiis the i-thtask, nis the number of tasks which are periodic, independent and preemptive, cidenotes the execution time of Ϣi, pidenotes the period of task Ϣi.

3.2. Control of frequency transmission

An approach to schedule a real time distributed system based upon modifications on frequency transmission of individual components in the system is presented in [4], this shows that scheduling of a distributed system can be accomplished through modifications on transmission frequencies into a region where the system performance is not affected. A linear time invariant model in which the coefficients of the state matrix are the relations between the transmission frequencies of each agent and through a feedback controller to modify transmission frequencies bounded between maximum and minimum values of transmission. This approach drives the frequency transmission based on three parameters: minimum frequency (fm), maximum frequency (fh)and real frequency (fr). Frequency transmission dynamics can be modeled as a linear time-invariant subsystem which state variables are transmission frequencies of the sensor nodes involved on the system. Note that for each primary task of a sensor Si, i=1,...,mfrequency can be expressed as fi=1/pi. There is a relationship between nodes’ frequencies an external input frequencies which serves as coefficients of the linear system. Therefore it is possible to control the NCS through the input vector usuch that the outputs yare nodes’ frequencies into a region Lbounded by maximum and minimum transmission frequencies, see Fig. 2.

Figure 2.

Transmission frequencies bounded by a schedulability region

The objective of controlling the frequency is to achieve coordination through the convergence of values. Each sensor Siknows its minimum and maximum frequencies based upon messages sent to controller and it could be estimated its own real transmission frequency. Let a NCS with a set χ=S1,S2,...,Slnodes that performs a set of task Ϣi=ci,pifor i=1,2,..,n, a subset of χis sensor nodes subset χs=S1,S2,...,Sm.

3.3. Network scheduling based on frequency transmission

An approach that modifies the frequency transmission uses fm, fr,fxfrequencies. RTDS dynamics, is modeled as a linear time-invariant system, whose state variables are frequencies transmission rates of the nnodes that compose the RTDS [4]. Frequency rates of a node are affected by some external input frequency rates, minimal frequencies of all nodes and particular ratios serve as coefficients of the linear system. So, it is possible to control the NCS using the input vector u, such that the output vector ycontains the frequency rates of all nodes within a nonlinear region L, bounded by the maximum and minimum transmission frequency rates. Let we assume that there is a relationship amongst real frequencies fr1,fr2,...,frmand external input frequencies u1,u2,...,umwhich serve as coefficients of the linear system:

xk+1=Axk+Buk,yk=Cxk.E1

A,B,Care matrices with appropriate dimensions. Ais the matrix of relationships between frequencies of sensor nodes, Bis the scale frequencies matrix, Cis the matrix with frequencies ordered, xk+1is a real frequencies vector in time t=k+1, ykis the vector of output frequencies. The input ukis a function of reference frequencies and real frequencies of the nodes in the distributed system, hence u=K(fm-fr)where fm,frare vectors. Then, the state vector in equation (1) can be written as:

rclxk+1=Axk+Buk,xk+1=Afrk+BK(fmk-frk).E2

Note that Kis the control gain defined as the basics of a LQR algorithm. Matrices A,Band Chas the proper dimensions and matrix Ainclude the following restriction [11]:

U=i=1nci/pi1,

as a new state of the system (1). For simplicity, frequency transition model will be described using a sensor node subset χs=S1,S2,S3,S4. The elements of the matrices for system 2 are defined as follows:

a¯ij=ιfm1,fm2,fm3,fm4fmii=jfmjfmiij,b¯ij=fhii=j0ij,c¯ij=1i=j0ij.

Note that ιfm1,fm2,fm3,fm4is the greatest common divisor of minimum frequencies, that is the planning cycle αexpressed in terms of frequencies of the backup task, for shortening, it will be written only as ι. fm, fh, and frare vectors of respective frequencies.

Considering the execution time ciof each task executing in respective sensor nodes in χsas an additional state, we can rewrite (2) as follows:

xk+11xk+12xk+13xk+14xk+15=ιfm1fm2fm1fm3fm1fm4fm10fm1fm2ιfm2fm3fm2fm4fm20fm1fm3fm2fm3ιfm3fm4fm30fm1fm4fm2fm4fm3fm4ιfm40c1c2c3c41fr1fr2fr3frnxc+fh100000fh200000fh300000fhn011111K100000K200000K300000K400000Kcfm1fm2fm3fm4xcr-fr1fr2fr3frnxc,yk1yk2yk3yk4yk5=1000001000001000001000001fr1fr2fr3fr4xc.xcis a real execution time and xcris a reference execution time.

Thus

xk+11xk+12xk+13xk+14xk+15=ιfm1fm2fm1fm3fm1fm4fm10fm1fm2ιfm2fm3fm2fm4fm20fm1fm3fm2fm3ιfm3fm4fm30fm1fm4fm2fm4fm3fm4ιfm40c1c2c3c41fr1fr2fr3frnxc+K1fh1(fm1-fr1)K2fh2(fm2-fr2)K3fh3(fm3-fr3)K4fh4(fm4-fr4)K1(fm1-fr1)+K2(fm2-fr2)+K3(fm3-fr3)+K4(fm4-fr4)+Kc(xcr-xr),yk1yk2yk3yk4yk5=1000001000001000001000001fr1fr2fr3fr4xc.

4. Real-Time Distributed System

This section exposes the RTDS used for implementation purposes, it is a 2-DOF helicopter prototype [13]. The following section briefly introduces and describes this 2-DOF helicopter prototype and its controller design.

The case of study is a prototype of a helicopter system integrated to a CanBus network with two propellers that are driven by DC motors. The front propeller controls the elevation of the helicopter nose about the pitch axis (ϖ) and the back propeller controls the side to side motions of the helicopter about the yaw axis (Ϧ). The pitch and yaw angles are measured using high-resolution encoders. A brief description of the helicopter model is presented, however detailed information can be found in [13]. The dynamics of the helicopter is developed based on kinetic and potential energy, this model is used to design a position controller. The helicopter centre of mass is described in xyzcartesian coordinates with respect the pitch and yaw angles, see Fig. 3.

Figure 3.

Dynamics of the 2DOF helicopter

The Euler-Lagrange equations are used to obtain nonlinear equations of motion for the 2 DOF Helicopter, which are used to derive the linear state model, and subsequently, to design the position controller. As the helicopter represents a non-linear system, it is required to perform a linearization around a point, this linearization point is

ϖ0=0,Ϧ0=0,ϖ̥0=0,Ϧ̥0=0.

From this, the linearization of the motion equation is obtained as follows:

Jeq,p+mhelilcm2ϖ̦=KppVm,p+KpyVm,y-Bpϖ̥-mheliglcm,E3
Jeq,y+mhelilcm2Ϧ̦=KppVm,y+KypVm,p-BpϦ̥-2mhelilcm2ϖϦ̥ϖ̥.E4

Substituing x=ϖ,Ϧ,ϖ̥,Ϧ̥'in (3) and (4) and solving for x̥the following linear model of state space is obtained:

x̥=0010000100-BpJeq,p+mhelilcm20000-ByJeq,y+mhelilcm2x+0000KppJeq,p+mhelilcm2KpyJeq,p+mhelilcm2-KypJeq,y+mhelilcm2-KyyJeq,y+mhelilcm2u,y=1000010000100001x.

where Kpp, Kyy, Kpy, Kypare the torque-constants used to obtain coupled torques acting on the pitch and yaw axes; for the state space model the input uand output yvectors are u=Vm,p,Vm,y'and y=x1,x2,x3,x4', Vmpis the input pitch motor voltage and Vmyis the input yaw motor voltage. Notice that, since the output matrix is the identity matrix, all states are measurable.

The model makes use of several Simulink and Matlab programs to develop the helicopter basic dynamics, by running a simulation of the closed-loop response, using the position controller. Regarding control issues, two controllers are designed: a FF-LQR and a FF+LQR+I. The FF+LQR regulates the pitch axis of the helicopter, using feed-forward (FF) and proportional-velocity (PV) compensators, while the yaw axis only makes use of a PV control. The FF+LQR+I controller uses an integrator in the feedback loop to reduce the steady-state error, by a feed-forward and proportional-integral-velocity (PIV) algorithms to regulate the pitch, and only a PIV to control the yaw angle. This work focuses on the FF+LQR+I controller, as follows. The FF+LQR control converges (ϖ,Ϧ,ϖ̥,Ϧ̥)(ϖd,Ϧd,ϖ̥d,Ϧ̥d)where ϖdis the desired pitch angle and Ϧdis the desired yaw angle, such that:

upuy=KffmheliglcmcosϖdKpp0.

The addition of an integrator requires to introduce the states x̥5=ϖand x̥6=Ϧ, so the linear state-space model is augmented as:

x̥=00100000010000-BpJeq,p+mhelilcm2000000-ByJeq,y+mhelilcm200100000010000x+0000KppJeq,p+mhelilcm2KpyJeq,p+mhelilcm2-KypJeq,y+mhelilcm2-KyyJeq,y+mhelilcm20000u,y=100000010000001000000100000010000001x.

Using the adequate Q and R weighting matrices, the control gain is as follows:

k=18.91.987.481.537.030.77-2.2219.4-0.4511.9-0.777.03.

Thus, the FF+LQR+I controller is:

upuy=KffmheliglcmcosϖdKpp0-k11k12k13k14k21k22k23k24ϖ-ϖdϦ-Ϧdϖ̥Ϧ̥-k15(ϖ-ϖd)+k16(Ϧ-Ϧd)k25(ϖ-ϖd)+k26(Ϧ-Ϧd).

5. Experimental approach

In order to study the impact of network utilization on closed control loop, the 2-DOF Helicopter control model is built as a NCS. Several nodes are connected through a common communication network. The experiment focuses on network scheduling, and the main objective is to balance the amount of data sent through the network, in order to avoid latency and under sampling.

The NCS for the experiment consists of 8 processors. These real-time kernel processors and the network are simulated using TrueTime [2, 12] based on Matlab/Simulink. The network used is a CSMA/AMP(CAN) with a transmission rate of 80000 bits/second, and not data loss. The NCS model is shown in Fig. 4.

Figure 4.

Networked Control System to analyze frequency transmission over the network

Four sensor nodes execute periodic tasks to sense control signals, as well as other additional periodic tasks. Each task has a period piand time consumption ci(Fig. 4). The sensed control signals are x=(ϖ,Ϧ,ϖ̥,Ϧ̥). This model has a controller node, depicted on the left side (Fig. 4). This controller takes the control law from the FF+LQR module by means of a task, which activates by event. The time consumption of the controller task is the maximum average time it takes to compute the control law. The controller node uses the values from sensors, and sends control outputs upand uy, that correspond to the pitch and yaw voltages. Two actuator nodes, located on the bottom right corner (Fig. 4), receive signals from the controller node. Finally, the scheduler node, located on the top right corner (Fig. 4), organizes the activity of the other seven nodes, and it is responsible for periodic allocation bandwidth, used by these nodes. Each node initializes, specifying the number of inputs and outputs of the respective TrueTime kernel block, defining a scheduling policy, and creating periodic tasks for the simulation. These tasks involve parameters about the periodic times and the consumption times. The task periodic times define the time interval between tasks, whereas the consumption times refer to the execution time of the task. Fig. 5 shows the 2-DOF Helicopter model, with a RTDS, where feedback control loop is closed through a communication network.

Figure 5.

Networked control integrated into closed control loop of the 2DOF Helicopter

Changes on the real-time task parameters of the RTDS commonly impact on network utilization, and therefore, on the control performance [8, 9]. The problem to tackle, thus, is to find a proper way to schedule the common communication network of the RTDS, based on managing an accurate sampling period, capable of keeping both, the network load and required integrated performance.

A criteria to quantify the system’s quality performance is the integral of the absolute value of the error (commonly expressed as IAE) is used:

IAE=t0tfe(t)dtk=k0kfr(kh)-y(kh)E5

where r(t)is reference signal or setpoint, y(t)is system output signal, t0(k0)and tf(kf)are the initial and final continuous(discrete) times of evaluation period [15].

5.1. Numerical simulations

In this section numerical simulations using the Network scheduling strategy based on Frequency Transmission are exposed. The network scheduling strategy dynamically adjusts the frequencies, considering the participation of several nodes of the NCS. These scheduling approach shows a way to manage the network resources, especially with a limited network bandwidth. These techniques avoid network delays during transmission. Numerical simulations shows that the dynamical changes of this strategy improve the RTDS response under fault scenarios.

The relationship amongst IAE and sampling periods of the primary sensing tasks is shown in Fig. (6). When the network transmits data without overloading or under sampling the output signals of pitch and yaw angles are similar to induced reference signal, so that the value of IAE is small. In contrast, when the transmission rate of sensor task set exceeds the upper bound data transmission rate the system becomes unstable, IAE increases accordingly.

Figure 6.

Values of the IAE using different sampling periods

The sensing periods was fixed in 10 miliseconds, sampling during the time simulation with this rate. In a fault-free scenario the pitch and yaw singnals track the reference signal, this behaviour is presented in Fig. 7

Figure 7.

Pitch and yaw signals in a fault-free scenario

It is assumed that in a fault scenario the change in periods of tasks may be produced, this induces a network utilization that possibily can not be supported by the bandwidth or perhaps stops data tranmission continuity. The effect of a fault over the RTDS is shown in the Fig. 8. In this case the periods of sensor tasks produce a low sampling resulting in loss of control.

Figure 8.

Pitch and yaw signals in a fault scenario

For the particular case study was made off-line analysis of the frequencies with which can be transmitted without increasing the value of the IAE for the sensing tasks. These ranges serve as parameters for the calculation of new transmission frequencies using the proposed model.

When fault occurs a signal is transmitted to the scheduler, this signal contains the id of sensor with fail and the parameters of the tasks on execution before the failure. The scheduler uses the frequency range and time of execution of the sensor tasks that has just entered, scheduler knows frequencies ranges of the tasks that continue without fail. This information feeds frequency transition model to compute a new frequency transmission range through a LQR control scheme, new frequencies are assigned to the sensor nodes in which the task period is changed therefore the effect of fault scenario is minimized balancing the amount of data sent over the network.

Numerical simulations were performed using the values of the maximum, minimum, and real frequencies besides the computational time, the values used for the sensor nodes are shown in the table 1.

NodeMax. Freq.Min. Freq.Real Freq.Consume
170280400.002
2502602500.002
3502501000.002
455300500.002

Table 1.

Maximum, minimum, and real frequencies besides the computational time

Figure 9 shows the frequencies controlled, this control bounds the frequency into a schedulability region where the IAE is low.

Figure 9.

Frequencies bounded into a schedulability region usign a frequency transmission controller

Figure 10.

Pitch and yaw signals in a fault scenario using frequency transmission model

Figure 10 shows system behavior during 30 seconds using frequency transition model. Nominal frequencies start in schedulability region during 8 secs, system transmits with this rate. In second 9 fault appears with a change of context, system transmits with this frequency until second 18. In second 19 frequency transition model change frequencies of backup tasks.

Figure 11 shows the computer network as well

Figure 11.

Sensing activiity and the related use of the computer networking using the Truetime toolbox

6. Conclusions

This work show a study of network scheduling strategy using numerical simulations. A simulation of a particular RTDS, a 2-DOF helicopter, is built using TrueTime as real-time simulation tool. A network scheduling strategy based on changes of frequency transmission rates is implemented in order to expose the advantages of using dynamic scheduling in an ad-hoc implementation for the network of a NCS. The use of numerical simulations aid to explore several considerations in design and analysis of a NCS.

© 2012 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution 3.0 License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

How to cite and reference

Link to this chapter Copy to clipboard

Cite this chapter Copy to clipboard

Oscar Esquivel-Flores, Héctor Benítez-Pérez and Jorge Ortega-Arjona (September 19th 2012). Issues on Communication Network Control System Based Upon Scheduling Strategy Using Numerical Simulations, Numerical Simulation - From Theory to Industry, Mykhaylo Andriychuk, IntechOpen, DOI: 10.5772/48578. Available from:

chapter statistics

1770total chapter downloads

More statistics for editors and authors

Login to your personal dashboard for more detailed statistics on your publications.

Access personal reporting

Related Content

This Book

Next chapter

On New High Order Iterative Schemes for Solving Initial Value Problems in Epidemiology

By Sandile Motsa and Stanford Shateyi

Related Book

First chapter

Modeling the Physical Phenomena Involved by Laser Beam – Substance Interaction

By Marian Pearsica, Stefan Nedelcu, Cristian-George Constantinescu, Constantin Strimbu, Marius Benta and Catalin Mihai

We are IntechOpen, the world's leading publisher of Open Access books. Built by scientists, for scientists. Our readership spans scientists, professors, researchers, librarians, and students, as well as business professionals. We share our knowledge and peer-reveiwed research papers with libraries, scientific and engineering societies, and also work with corporate R&D departments and government entities.

More about us