Systematic Approach for Scheduling of Tasks and Messages under Noise Environment

Similar documents
Performance Analysis of an Enhanced DQRUMA/MC-CDMA Protocol with an LPRA Scheme for Voice Traffic

A Non-cooperative Game Theoretic Approach for Multi-cell OFDM Power Allocation Ali Elyasi Gorji 1, Bahman Abolhassani 2 and Kiamars Honardar 3 +

LMS Beamforming Using Pre and Post-FFT Processing for OFDM Communication Systems

A Cooperative Spectrum Sensing Scheme Based on Trust and Fuzzy Logic for Cognitive Radio Sensor Networks

986 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 33, NO. 5, MAY 2015

Definition of level and attenuation in telephone networks

Redes de Comunicação em Ambientes Industriais Aula 8

User Based Resource Scheduling for Heterogeneous Traffic in the Downlink of OFDM Systems

Resource Scheduling in Dependable Integrated Modular Avionics

Digital Transmission

LS-SVM Based WSN Location Algorithm in NLOS Environments

Priority based Dynamic Multiple Robot Path Planning

Comparative Analysis of Reuse 1 and 3 in Cellular Network Based On SIR Distribution and Rate

Optimal Placement of Sectionalizing Switches in Radial Distribution Systems by a Genetic Algorithm

The Synthesis of Dependable Communication Networks for Automotive Systems

Neuro-Fuzzy Network for Adaptive Channel Equalization

Topology Control for C-RAN Architecture Based on Complex Network

UWB & UWB Channels HANI MEHRPOUYAN

TODAY S wireless networks are characterized as a static

Dynamic Resource Control for High-Speed Downlink Packet Access Wireless Channel

Resource Allocation Optimization for Device-to- Device Communication Underlaying Cellular Networks

Performance Analysis of MIMO SFBC CI-COFDM System against the Nonlinear Distortion and Narrowband Interference


AN ALGORITHM TO COMBINE LINK ADAPTATION AND TRANSMIT POWER CONTROL IN HIPERLAN TYPE 2

antenna antenna (4.139)

THE third Generation Partnership Project (3GPP) has finalized

Cooperative Wireless Multicast: Performance Analysis and Power/Location Optimization

TECHNICAL NOTE TERMINATION FOR POINT- TO-POINT SYSTEMS TN TERMINATON FOR POINT-TO-POINT SYSTEMS. Zo = L C. ω - angular frequency = 2πf

Distributed Adaptive Channel Allocation in Multi-Radio Wireless Sensor Networks

Recurrent Neural Network Based Fuzzy Inference System for Identification and Control of Dynamic Plants

Exploiting Dynamic Workload Variation in Low Energy Preemptive Task Scheduling

Walsh Function Based Synthesis Method of PWM Pattern for Full-Bridge Inverter

A ph mesh refinement method for optimal control

Modeling Hierarchical Event Streams in System Level Performance Analysis

A study of turbo codes for multilevel modulations in Gaussian and mobile channels

Evaluation of Kolmogorov - Smirnov Test and Energy Detector Techniques for Cooperative Spectrum Sensing in Real Channel Conditions

Calculation of the received voltage due to the radiation from multiple co-frequency sources

Optimal and water-filling Algorithm approach for power Allocation in OFDM Based Cognitive Radio System

Optimized Forwarding for Wireless Sensor Networks by Fuzzy Inference System

Multi-Source Power System LFC Using the Fractional Order PID Controller Based on SSO Algorithm Including Redox Flow Batteries and SMES

High Speed, Low Power And Area Efficient Carry-Select Adder

A Tractable and Accurate Cross-Layer Model for Multi-Hop MIMO Ad Hoc Networks

Sensors for Motion and Position Measurement

Prevention of Sequential Message Loss in CAN Systems

Practical Issues with the Timing Analysis of the Controller Area Network

Performance Analysis of Multi User MIMO System with Block-Diagonalization Precoding Scheme

Adaptive Modulation for Multiple Antenna Channels

MTBF PREDICTION REPORT

COMPARISON OF DIFFERENT BROADCAST SCHEMES FOR MULTI-HOP WIRELESS SENSOR NETWORKS 1

Keywords LTE, Uplink, Power Control, Fractional Power Control.

To: Professor Avitabile Date: February 4, 2003 From: Mechanical Student Subject: Experiment #1 Numerical Methods Using Excel

Rejection of PSK Interference in DS-SS/PSK System Using Adaptive Transversal Filter with Conditional Response Recalculation

A Multi-standard Efficient Column-layered LDPC Decoder for Software Defined Radio on GPUs

Uncertainty in measurements of power and energy on power networks

Secure Transmission of Sensitive data using multiple channels

IEE Electronics Letters, vol 34, no 17, August 1998, pp ESTIMATING STARTING POINT OF CONDUCTION OF CMOS GATES

Network Reconfiguration in Distribution Systems Using a Modified TS Algorithm

A Digital Content Distribution Using a Group-Key and Multi-layered Structure Based on Web

Multiarea Transmission Cost Allocation in Large Power Systems Using the Nodal Pricing Control Approach

Queuing-Based Dynamic Channel Selection for Heterogeneous Multimedia Applications over Cognitive Radio Networks

Non Pre-emptive Scheduling of Messages on SMTV Token-Passing Networks

Distributed Channel Allocation Algorithm with Power Control

Radial distribution systems reconfiguration considering power losses cost and damage cost due to power supply interruption of consumers

Low-Complexity Factor Graph Receivers for Spectrally Efficient MIMO-IDMA

Dynamic Optimization. Assignment 1. Sasanka Nagavalli January 29, 2013 Robotics Institute Carnegie Mellon University

A Tractable and Accurate Cross-Layer Model for Multi-Hop MIMO Networks

Throughput Maximization by Adaptive Threshold Adjustment for AMC Systems

A Mathematical Model for Restoration Problem in Smart Grids Incorporating Load Shedding Concept

Equivalent Line Limit Calculation Using Available Thermal Transfer Capability

Graph Method for Solving Switched Capacitors Circuits

A Fuzzy-based Routing Strategy for Multihop Cognitive Radio Networks

Full-Duplex Device-to-Device Collaboration for Low-Latency Wireless Video Distribution

Dynamic SON-Enabled Location Management in LTE Networks

Adaptive Modulation and Coding for Utility Enhancement in VMIMO WSN Using Game Theory

Passive Filters. References: Barbow (pp ), Hayes & Horowitz (pp 32-60), Rizzoni (Chap. 6)

Revision of Lecture Twenty-One

6928 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 13, NO. 12, DECEMBER 2014

A New Regressor for Bandwidth Calculation of a Rectangular Microstrip Antenna

Distributed Computation in Dynamic Networks

Throughput Analysis Model for IEEE e EDCA with Multiple Access Categories

Integrity Data Attacks in Power Market Operations

NOVEL ITERATIVE TECHNIQUES FOR RADAR TARGET DISCRIMINATION

Define Y = # of mobiles from M total mobiles that have an adequate link. Measure of average portion of mobiles allocated a link of adequate quality.

OPTIMAL GUARANTEED SERVICES TIMED TOKEN (OGSTT) MEDIA ACCESS CONTROL (MAC) PROTOCOL FOR NETWORKS THAT

Power Minimization Under Constant Throughput Constraint in Wireless Networks with Beamforming

On Interference Alignment for Multi-hop MIMO Networks

A High-Sensitivity Oversampling Digital Signal Detection Technique for CMOS Image Sensors Using Non-destructive Intermediate High-Speed Readout Mode

Efficient Power Allocation for LDPC-Coded MIMO Systems

Multi-objective Transmission Planning Paper

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY 1

ESTIMATION OF DIVERGENCES IN PRECAST CONSTRUCTIONS USING GEODETIC CONTROL NETWORKS

Phasor Representation of Sinusoidal Signals

High Speed ADC Sampling Transients

A MODIFIED DIRECTIONAL FREQUENCY REUSE PLAN BASED ON CHANNEL ALTERNATION AND ROTATION

A Data-Driven Robustness Algorithm for the Internet of Things in Smart Cities

The Spectrum Sharing in Cognitive Radio Networks Based on Competitive Price Game

Power-Constrained Test Scheduling for Multi-Clock Domain SoCs

Figure 1. DC-DC Boost Converter

Chapter 2 Two-Degree-of-Freedom PID Controllers Structures

Journal of Applied Science and Agriculture, 9(4) April 2014, Pages: Journal of Applied Science and Agriculture

Transcription:

Systematc Approach for Schedung of asks and Messages under Nose nvronment Hyoung Yuk KIM Hye Mn SHIN and Hong Seong PARK Dept of ectrca and omputer ng Kangwon Natona Unversty 9- Hyoja Dong huncheon 00-70 Korea Abstract Hgh degree of MI (ectro-magnetc Interference) or nose s generated n the pant envronment where contro systems consst of smart sensors smart actuators and controers connected va a fedbus he nose generated by some devces such as hgh-power motors may cause communcaton errors and deay the successfu transmsson of data herefore the nose condton s one of the mportant parameters consdered n the desgn of a reabe network-based contro system hs paper presents the schedung method of task and message to guarantee the gven end-to-end constrants under nose envronments he presented schedung method s apped to an exampe of a contro system that uses AN (ontroer Area Network) consderng two knds of nose modes he comparson resuts for each nose condton shows the mportance of consderng the nose condton n system desgn System desgners are abe to desgn the contro system guaranteeng ts requrements under a nose envronment by usng the proposed schedung method Keywords: Nose Rea-me Schedung Fedbus INRODUION Nowadays contro systems generay consst of smart sensors and smart actuators whch are connected wth controers va fedbuses such as Profbus FIP AN and LonWorks Because sensors actuators and controers may be dstrbuted data transmsson from sensor to controer and from controer to actuator may be randomy deayed due to network characterstcs Aso the task executon tme at a node depends on the node s characterstcs such as the operatng system the number of ts tasks and so on herefore the network-nduced random deay and the varabe task executon tme shoud be consdered n system desgn [-] For exampe f task A for transmttng a message to another node has ower prorty than task B for dagnostcs at the same node but starts smutaneousy the former s competed after the atter s fnshed hs way the worst-case response tme s vared accordng to the task s prorty In addton message transmsson can be deayed f ts prorty s ower than the prortes of other messages that are competng to use the avaabe meda In other words the worst-case response tmes depend on the prortes of tasks and messages here are severa researches on the system schedung method for sovng probems mentoned above [4-8] A heurstc schedung method s proposed n [4] whch assumes that any one task cannot receve more than one message A schedung method usng severa prunng agorthms s proposed n [6] whch extends a task-based schedung method [5] for dstrbuted contro systems In [7] a method for schedung dstrbuted systems consstng of SMA/A and DMA networks s proposed hs method uses a genetc agorthm and a custerng agorthm In [8] a perod assgnment method for tasks and a heurstc assgnment rue for message's prorty are proposed and the schedung method whch uses the prevous two methods for a practca dstrbuted contro system consstng of mutpe contro oops s proposed he above research assumed error-free or nose-free networks In other words they cannot be apped to pants under nose envronments Most dstrbuted contro systems are nstaed and operated n the pant wth generay many hgh-power motors herefore hgh degrees of ectro Magnetc Interferences (MIs) from these motors strongy affect the contro systems [90] and these nterferences may cause the transmsson errors of data he transmsson errors may ead to performance degradaton or serous fauts n the contro systems because they make the successfu transmsson of data deayed For exampe f the end-to-end response tme of a contro oop ncreases due to the transmsson errors the contro system mght mss the contro oop s deadne herefore when a dstrbuted contro system operated under nose envronment s desgned the transmsson error due to nose shoud be consdered to satsfy gven reatme constrants herefore a new method for desgnng dstrbuted contro systems under nose envronment s requred here s some research on message transmsson deay caused by errors An anayss method on the effect of nose s presented n [] It uses an error mode based on the bt error rate of smpe sporadc nose and presents the anayss method of a message s response tme transmtted va AN under ths nose mode Aso a method that can consder mutpe nose sources smutaneousy s presented n [9] But these research however have anayzed ony the message transmsson deay affected by the transmsson errors and have not consdered the task executon deay caused by precedence reatons between tasks and messages and the end-to-end deay from sensors to actuators caused by transmsson errors hs paper proposes the schedung method of task and message for contro systems operatng under nose envronment and verfes the proposed schedung method by appyng t to an exampe of a dstrbuted contro system connected va AN he proposed method schedues the perods and the prortes of SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6 9

tasks and messages to satsfy the gven end-to-end constrants under nose envronment and extends the schedung method proposed n [8] to consder the communcaton errors caused by nose We used two knds of nose modes adoptng mutsources for the nose he schedung resuts namey the system s end-to-end response tmes and the messages response tmes are derved for each nose mode In Secton a nose mode adoptng mut-nose sources and an anayss method for the message response tmes are descrbed he proposed schedung method s apped to the exampe system and the anayss of the schedung resuts s presented n Secton Fnay concusons are gven n Secton 4 NOIS MODL AND DLAY IM Probems caused by nose Most contro systems operate n pants usng hgh-power motors In ths envronment hgh degrees of MI may generate communcaton errors hese errors deay the message transmsson between nodes and may cause crtca fauts n the contro systems hs probem s ustrated n Fg In ths fgure there are two contro oops: he st contro oop conssts of Node Node ontroer and ActuatorNode; the nd contro oop conssts of Node Node ontroer and ActuatorNode ach node has ony one task A tasks are nvoked smutaneousy and contro tasks and actuator tasks wat unt the nput messages are receved he messages are prortzed as foows: ask foowed by ask ask ontroask and ontroask It s assumed that the deadne of each contro oop s equa to each contro task s perod sampng to actuator outputtng are satsfed wthn the deadne of each contro oop when there s no nose If nose causes errors at that tme the ask s message s sent an error recovery mechansm such as retransmsson method as shown n Fg can correct the message transmsson Because of ths communcaton deay ontroask and ontroask whch wat for a message from ask are deayed Actuator ask and Actuatorask are aso deayed Fnay the end-toend response tme of the nd contro oop becomes onger than ts deadne hs ncrease n response tme ndcates the occurrence of a crtca faut n the contro system herefore communcaton errors caused by nose shoud be consdered n desgn of dstrbuted contro system operatng under nose envronment to guarantee the rea-tme requrements Nose mode and deay tme [9] o consder nose n system desgn a proper nose mode whch can anayze the effects of nose on the system shoud be consdered We used a nose mode that had mutpe nose sources [9] as shown n Fg b I n n b = 4 n = Burst Nose Fg Nose mode 4 b Resdua Nose r I r t Node Node Node ontroer ontroer Actuator Node Actuator Node Fedbus : ask omputaton : Message ransmsson Perod end-to-end response tme : Nose : ask's Invocaton Successfu Retransmsson Fg nd-to-end response tme deayed by nose Deadne Mss As shown n Fg the end-to-end constrants from sensor Fg represents one nose source whch conssts of a burst nose group and a resdua nose group he burst nose group s composed of b subgroups beng generated wth the mnmum perod b and each subgroup conssts of n noses occurrng wth the mnmum perod n durng I n he resdua noses occur wth the perod r durng I r after the burst nose Usng ths nose mode communcaton deay tme can be derved from error detecton and recovery mechansm of the correspondng network If a nose whch can be represented by ths mode s generated we can get the number of noses occurrng for the perod t by cacuatng the number of burst noses Bu (t) and that of resdua noses Re (t) separatey If the perod t ncudes the part of the resdua nose Bu (t) s equa to n * b Otherwse Bu (t) s equa to the sum of the number of noses of the subgroups whoy ncuded n the perod t tb n and that of noses of the subgroups party ncuded n the perod t tmod b n herefore Bu (t) s equa to () t tmod b Bu ( t) = mn n b n + mn n b n () 94 SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6

If the perod t ncudes ony the part of burst noses the number of resdua noses Re(t) s 0 If the perod t ncudes the part of resdua noses we can get Re(t) as foows: t b b Re ( t) = max 0 r Let (t) be the transmsson deay tme of message caused by tota number of noses durng perod t and (t) be that of message caused by nose source durng the perod t If there are k nose sources () t s equa to () k () t = () t + () t + + () t () If communcaton errors occur n the AN the worst-case response tme of message can be cacuated by the presented nose mode and () ~ () he overhead of message O caused by one error bt n the AN conssts of error recovery bts and the message retransmsson [67] and s represented as (4) O = + max ( ) (4) bt k k hp() {} bt s the tme for transmttng one bt and hp() s a set consstng of messages wth hgher prorty than message k s the rea transmsson tme of message In order to evauate n the worst-case the maxmum transmsson tme between correspondng messages s used Usng (4) (t) s the product of the overhead and the number of error bts by burst noses and resdua noses If nose duraton of burst nose part I b and that of resdua one I r are onger than one bt tme the dfferences shoud be compensated herefore (t) s equa to (5) () t = Bu () t ( O + max(0 Ib bt)) + Re ( t) ( O + max(0 I )) r bt he transmsson deay of message s nfuenced by a noses occurrng from the tme that the message s nserted nto the transmsson queue of AN to the tme that the message s transmtted competey herefore the transmsson deay by nose s ncuded n the queung deay of message q as shown n (6) q + bt + J q = B + j + ( q + ) (6) j hp() j B s the maxmum bockng tme for whch the message can be deayed by the messages wth ower prorty than ts prorty J s a jtter that occurs when message s nserted nto the transmsson queue of AN and j s the perod of message In the AN the worst-case response tme of message R - from the tme that the message s nserted nto the transmsson queue of AN to the tme that the message s transmtted competey conssts of jtter for nsertng nto the queue J the queung deay q and the message transmsson tme and s shown n (7) () (5) he worst-case response tme of message R under nose condton shoud be equa or ess than the message s deadne d herefore the reaton R d s estabshed and R can be used as the predctabe deadne of message SHDULING OF ASKS AND MSSAGS UNDR NOIS NVIROMN ask and message schedung Because the prorty and the perod of message and task have a very mportant effect on the rea-tme characterstcs of dstrbuted contro systems system schedung must assgn the prorty and the perod guaranteeng the system s rea-tme requrements n the system desgn he schedung process proposed n [8] conssts of four steps and s shown n Fg In step a task graph s desgned accordng to a system mode and ts gven requrements and then each task s aocated to a physca node In step the equatons or the nequates of constrants on tasks and messages are derved from the task graph In step a prorty and a perod are aocated to each task and message by the prorty and perod assgnment agorthm In step 4 the derved equatons or nequates are soved wth the worst-case response tmes of messages and tasks If the soved resuts do not meet the end-to-end constrants ths method goes back to step and changes some of the parameters of tasks and messages such as perod or prorty hs process s repeated unt the cacuated resuts satsfy the end-to-end constrants If the perod and the prorty that meet the constrants cannot be found the schedung process returns to step and changes the system s requrements or redesgns the task graph he schedung method presented above s extended to consder the nose condton n the system schedung and the extended method s shown n Fg he addtona processes are the nose modeng of an actua spot at the tme of system requrement dervaton and the dervaton of deay tme equatons due to nose at the tme of constrant dervaton Fnay the derved deay tme equatons are used to sove derved constrants after the perods and the prortes are assgned Wth ths method one can schedue the system consderng nose condton of the spot Because nodes or statons can be dstrbuted wthn the dstrbuted contro systems nose characterstcs vary accordng to where these are nstaed For exampe f nodes or communcaton meda are nstaed near hgh power motors these nodes or meda may be nfuenced by hgh power nose But f they are nstaed at a pace wthout a nose source ther error rate may be very ow or zero herefore f nose condtons can be apped dfferenty dependng on the spots systems can be desgned more sutaby But ths paper eaves ths consderaton to future works and assumes nose condtons are same wthn the systems R = J + q + (7) SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6 95

hange Requrements No Yes Redesgn? No Yes Reassgn? No Yes Reassgn? nd-to-nd Requrements / Derve Nose Mode ask Graph Desgn / ask Aocaton Derve onstrants / Derve Nose Anayss q Prorty Assgnment Perod Assgnment Step Step Step Node 4 ontroer and ActuatorNode; and the ast contro oop conssts of Node4 5 6 ontroer and ActuatorNode4 5 A nodes are connected va AN LOOP ontroer Actuator Actuator Node Node ontroer Actuator Node ontroer Node Node LOOP A A A AN Node Node4 Node5 LOOP Node6 A4 Actuator Node4 Actuator Node5 No Sovng onstrants usng Nose Anayss q Meet onstrants? Yes Startng tmes Deadnes Prortes Perods of asks and Messages Step 4 Fg ask and message schedung under nose envronment Notatons he foowng notatons w be used throughout ths paper α the task α ncuded n the -th contro oop (=~) and α {Sj Ak m Om } where Sj Ak m and Om denote the j-th sensor task (j=~6) a contro task the k-th actuator task (k=~5) the -th message task (=~) and the m-other task extng n a contro node but not beng reated to contro oop (m=~) respectvey j α the task α that s ncuded n the -th contro oop and the j-th contro oop at the same tme and α s equa to the defnton of α e α the executon tme of α α the perod of α d α the deadne of α p α the prorty of α α the nta phase tme or startng tme of α π β the port for the message β ncuded n the -th contro oop and β {Sj k m } where Sj k and m denote the message of the j-th sensor task (j=~6) the k-th message of a contro task (k=) and the message of the -th message task (=~) respectvey π j β the port for the message β that s ncuded n the -th contro oop and the j-th contro oop at the same tme and β s equa to the defnton of π β MAD the end-to-end maxmum aowabe deay tme of the - th contro oop System mode and task graph An exampe system to whch the proposed schedung method s apped and ts task graph are shown n Fg4 and Fg5 respectvey he system conssts of 6 sensor nodes controers and 5 actuator nodes that form three contro oops he st contro oop s composed of Node 4 ontroer and ActuatorNode ; the nd contro oop s S S S S4 S5 S6 S S S S4 Nodes S S S S4 S 5 S 6 π S π S π S π S4 π S 5 π S 6 Network m m m m4 m5 m6 Fg4 arget System π m π m π m π m4 π m5 π m6 ontroers O O O O O O O O O π π π π π S5 Network m7 m8 m9 m0 m Fg5 ask graph of target system S6 π m7 π m8 π 9 π m0 π A A A5 Actuator Nodes m A A4 A5 he parameters of system requrements prevousy gven before schedung the system are summarzed n abe here are the end-to-end deadnes of three oops the executon tme of each task the sze of each message and the nformaton of the tasks that exst n the controers but are not ncuded n the contro oops abe System parameters In abe two knds of nose modes-as and AS-to be apped to the target system are represented usng the method presented n Secton In case of AS two nose sources exst smutaneousy he operatona A A A A4 A5 96 SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6

characterstcs of the target system under two knds of nose condtons were anayzed and compared he resuts showed that the proposed schedung method coud schedue the system under these nose condtons Aso the compared schedung resuts showed that the effect of nose was mportant to contro systems abe rror modes 4 Dervng onstrants After dervng the system requrements and the nose mode the system s task graph s drawn and each task s assgned to each node And then equatons or nequates representng the system requrements and the reaton of task and message are derved from the task graph he derved equatons or nequates are used as constrants at the system schedung onstrants consst of one for the end-to-end response tme of contro oops one for the perod reatons between tasks and messages and one for the precedence reatons between tasks and messages he constrants for the st contro oop are as foows: he end-to-end tme constrant s max( A + da A + da) MAD he perod constrants of tasks and messages are S = m S = m S = m S4 = m4 m m m m4 = m7 = A = m8 = A he precedence constrants of tasks and messages are S+ ds m S + ds m S + ds m S4 + ds4 m4 max( m + dm m + dm m + dm) + d m7 + d m8 m7 + dm7 A m8 + dm8 A he constrants about the nd contro oop are as foows: he end-to-end tme constrant s A + da MAD he perod constrants of tasks and messages are S = m S4 = m4 m m4 = m9 = A he precedence constrants of tasks and messages are S + ds m S4 + ds4 m4 max( m + dm m4 + dm4 ) + d m9 m9 + dm8 A he constrants for the rd contro oop as foows: he end-to-end tme constrant s max( A4 + da4 A5 + da5) MAD he perod constrants of tasks and messages are S4 = m4 S5 = m5 S6 = m6 m4 m5 m6 = m 0 = A4 = m = A5 he precedence constrants of tasks and messages are S4 + ds4 m4 S5 + ds5 m5 S6 + ds6 m6 max( m4 + dm4 m5 + dm5 m6 + dm6) + d m 0 + d m m0 + dm 0 A4 m + dm A5 5 Schedung resuts o schedue the system by usng the constrants derved n subsecton 4 the prorty and the perod of task and message was assgned by usng the assgnment agorthms proposed n [8] to assgn these After assgnng the worst-case response tmes of tasks and messages were cacuated by the anayss method descrbed n subsecton he cacuated response tme of task and message were substtuted for the deadnes of the derved constrants and then the derved constrants were soved If the resuts met the nequates of end-to-end tme constrants the parameters of task and message such as prorty perod and nta phase tme were saved as scheduabe parameters he schedung resuts are ustrated from Fg6 to Fg0 hey represent the perods and the end-to-end worst-case response tmes of contro oops n the cases of two knds of nose modes and no nose condton he X-axs represents the number of the teratve cacuatons of the perod assgnment agorthm n [8] and the Y-axs represents the end-to-end response tme and perod of contro oops he perod assgnment agorthm tres to fnd out the smaest perod of the contro oop that aso sustans the system to meet ts rea-tme requrements hat s the perod of contro oop has to be onger than the end-to-end worst-case response tme hs agorthm repeats a searchng agorthm based on a bsecton agorthm unt the smaest one s founded Fg6 AN: 00kbps under no nose condton SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6 97

In Fg6 the assgned perod and the end-to-end worst-case response tme of each contro for no nose source and 00kbps AN are represented Fg6 shows that the perods of the contro oops are set to 0ms 0ms and 45ms and the end-to-end response tmes are 5ms 6ms and 6ms hese schedung resuts mean that the system can be schedued by settng the perods of the contro oops to 0ms 0ms and 45ms under these condtons of perods for the no nose condton and the AS nose condton but cannot fnd one for the AS nose condton In order to schedue the system for the AS nose condton we shoud change some parameters such as the network speed For exampe et us change the network speed to 500 kbps When 500kbps AN s used the schedung resuts for AS and AS nose condtons are shown n Fg9 and Fg0 Fg7 AN: 00kbps under AS nose condton Fg 9 AN: 500kbps under AS nose condton In Fg7 the assgned perod and the end-t-end worst-case response tme of each contro are represented when AS nose condton of abe was apped to the system and the speed of AN was 00kbps he perods of contro oops were 60ms 80ms and 00ms and the end-to-end response tmes were 50ms 50ms and 70ms herefore the system can be schedued under these condtons Fg0 AN: 500kbps under AS nose condton Fg8 AN: 00kbps under AS nose condton Fg8 shows the schedung resuts obtaned when AS nose condton of abe havng two nose sources was apped to the system and the speed of AN was 00kbps Because of hgh degrees of nose the transmsson deay of each message ncreased and ths deay ncreased the end-to-end response tme of the contro oop Fg6 Fg7 and Fg8 show that we can fnd a feasbe set Fg9 and Fg0 show that the system can be schedued for AS nose condton as we as AS Despte the AS nose condton the system can be schedued f the perods are set to 0ms 5ms and 40ms and the response tmes to 9ms 0ms and 5ms From these resuts we know that the system requrements must be changed to guarantee the performance accordng to nose condton Usng the proposed method we can evauate and desgn the system to meet ts requrements under nose condtons before run-tme he worst-case response tmes of a messages are depcted accordng to the network speed and the nose modes n Fg At 00kbps the response tmes of the messages vared markedy accordng to the three nose condtons hs varaton means that t takes much tme to recover the errors because 00kbps s a reatvey ow speed At 500kbps the response tmes vared ess than that at 00kbps Aso the response tmes 98 SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6

Fg Worst-case response tmes of a messages of 500kbs under AS nose condton s smar to those of 00kbs under no nose condton ven though the deay caused by error may be reatvey sma to the response tme of each message the transmsson errors have a heavy effect on the system because the end-to-end response tme of a contro oop s the sum of a response tmes of the messages A hgher transmsson speed s one good souton to sovng the schedung probem caused by the transmsson errors When AN network speed s 500kbps and nose condton s that of AS the deta schedung resuts are shown n abe he parameters of message and task such as the prorty the perod the nta phase tme and the deadne were cacuated usng the proposed method By settng the statc parameters of the system to these the system can guarantee ts performance under the nose envronment at run-tme abe AN: 500kbps under AS condton 4 ONLUSIONS Most dstrbuted contro systems operate n the pant generay havng many devces such as hgh power motor whch generate nose hese motors generate hgh degrees of ectro Magnetc Interferences (MIs) hese nterferences have much effect on the contro system and ts network and cause transmsson errors whch ead to performance degradaton such as onger transmsson deay or serous fauts n the contro systems herefore the transmsson error due to nose shoud be consdered when tryng to satsfy the gven reatme constrants n the system desgn hs paper proposed the schedung method of both task and message for contro systems under nose envronment and the proposed method was verfed through an exampe wo knds of nose modes were apped to AN and the end-to-end response tme of contro oop and the response tme of message were anayzed he resuts showed that the transmsson deay caused by the transmsson error s crtca to system performance and that the system requrements must be changed dependng on the nose condtons Usng the proposed schedung method system desgners can desgn a system guaranteeng ts performance whe the system operates under nose envronment Future works shoud focus on the dervaton method for the nose mode of an actve spot and the appcaton method for the nose condtons accordng to the pace wthn a system RFRNS [] A Burns Preemptve prorty based schedung: An approprate engneerng approach n Prncpes of Rea-me Systems Prentce Ha 994 [] J Xu and D Parnas "Schedung processes wth reease tmes deadnes precedence and excuson reatons" I r on Software ngneerng pp 60-69 March 990 [] J Y- Leung and J Whtehead "On ompexty of Fxed- Prorty Schedung of Perodc Rea-me asks" Performance vauaton (4) pp 7-50 December 98 [4] K nde "Hostc Scheduabty Anayss for Dstrbuted Hard Rea-tme Systems" echnca Report YS-97 Dept of omputer Scence Unv of York NOV 994 [5] R Gerber and SS Hong "Guaranteeng Rea- me Requrements wth Resource-Based abraton of Perodc Processes" I r on Software ngneerng (7) Juy 995 [6] JW Park YS Km SS Hong M Saksena SH Noh and WH Kwon "Network conscous of Dstrbuted Rea- me Systems" Journa of System Archtecture pp -56 998 [7] S Faucou A-M Depanche and J-P Beauvas "Heurstc echnques for Aocatng and Schedung ommuncatng" WFS-000 pp 57-65 000 [8] HY Km M Lee and HS Park "Optma Perod and Prorty Assgnment Usng ask & Message-Based Schedung n Dstrbuted ontro Systems" Journa of ontro Automaton and Systems ngneerng Vo8 No6 JUN 00 [9] Saskumar Punnekkat Hans Hansson and hrster Norston "Response me Anayss under rrors for AN" Proc of the 6th Rea-me echnoogy and SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6 99

Appcatons Symposum pp58-65 MAY 000 [0] Lang a and We Zhang "MI n hydropower pant and M desgn for ts computer montorng and contro system" the rd Internatona Symposum on ectromagnetc ompatbty pp78-8 00 [] K nde and A Burns "Guaranteed Message Latences for Dstrbuted Safety rtca Hard Rea-me Networks" YS 9 Dept of omputer Scence Unv of York June 994 [] AN n Automaton (A) AN Specfcaton 0 Part A and Part B http://wwwcan-cade [] Inte 857 Sera ommuncatons ontroer Archtectura Overvew Jan 996 [4] K nde H Hansson and A Wengs "Anayzng Reatme ommuncatons: ontroer Area Network" I Rea-tme Systems Symposum 994 00 SYSMIS YBRNIS AND INFORMAIS VOLUM - NUMBR 6