Presentation of a new Local Area Network for C.I.M.

applications

A. Kalogeras, K. Efstathiou, G. Zapantes, S. Koubias, G. Papadopoulos

Applied Electronics Laboratory, Department of Electrical Engineering

Abstract- Computer Integrated Manuf'acturiug may solve a lot Intel BITBUS is a hierarchical centralised network
of problems modern industry faces, by integrating the consisting of one master and multiple slaves. The objective
computer in all manufkctnrlng processes This paper presents of the BITBUS interconnect is to provide an interface
an industrial Local Area Network that has being developed between master and slave node tasks. This interface is
under the h.amewor% of the European STRIDE LIGHT
project placing greater emphasis on the network management accomplished via an order and reply message protocol. At
that has been implemented. the same time the BITBUS Application Layer provides a
subset of the Manufacturing Message Specification ( M M S )
171.
I. INTRODUCTION The paper at frrst presents the idea of the industrial
network used and then analyses the management structure
that has been implemented. It must be mentioned that the
Computer Integrated Manufacturing has brought a whole system was developed under the auspices of the
revolution in the way we think about industrial applications European STRIDE LIGHT project [8]. It must be also
and their architecture. While the traditional architecture of noted that the presented system has been installed and is
an industrial application is based on parallel channels working successfully in five collaborating big Greek
(buses)which interconnect clusters of censors and actuators, industries.
C.I.M. proposes an architecture using a multi-point
industrial network. This results in major improvements, not
the least important among which are the significant cuts in 11. STRUCTUREOF THEINDUSTRIAL NETWORKTHATHAS
cabling, design, installation and adjustment costs, the ease BEEN IMPLEMENTED
in maintenance and modification, a very important increase
in the system reliability, the easier and wider accessibility to
variables, the ability to have a central database that means As mentioned in the introduction, Intel BITBUS (IEEE
ability to have various reports as well as visualise a given std. 1118) is a centralised hierarchical local area industrial
application and finally the ability to have advanced control network. There is one master node and several slave nodes.
on system data. The transfer of data via the network takes place using an
All these improvements have resulted in making the order and reply message protocol. In addition a subset of
installation of a local area network in an industrial the Manufacturing Message Specification is provided.
application a requirement rather than a characteristic of Each slave node is based on an industrial controller
C.I.M. The general idea of a local area network for an which was built round the powerful 8344 Intel
industrial application is that of a spatially distributed data Microcontroller [9]. This microcontroller encompasses the
base, which data have to be transferred via the network in 8051 Intel microcontroller and a Serial Interface Unit (SIU)
regular intervals. The data transfer that takes place is based which provides network facilities according to the IBM
on a master slave scheme. Several fieldbus networks may. SDLC format. Both the Physical and the Data Link Layer
be used like MINIMAP, FIP, PROFIBUS, Intel BITBUS, of the OS1 Reference Model are implemented by hardware
e.t.a. [l], [2], [3], [4], [5]. This paper describes a system on the chip. Fwthermore a real time operating environment
based on the Intel BITBUS network. named iDCX is adopted in order to use the optimum of the
The Intel BITBUS is defined with respect to a protocol computing capacities of the chip [lo].
implementing only three of the seven layers of the OS1 The selection of a powerful microcontroller like the
Reference Model with a vertical network management layer. Intel 8344 as the microcontroller to support the industrial
The three layers are the Physical (Layer 1), the Data Link controller is a successful one because this chip has to
(Layer 2) and the Application Layer (Layer 7). [6] The execute in parallel both the sufficient and reliable exchange
of data and the execution of preprocessings on the

532
 Fig.1. Global Memory Access Scheme

controller. These preprocessings are about tasks that reside It provides the system with the ability to implement
on the network node and serve the needs of the given periodical polling of the 'global memory areas' of the
application. nodes wig.1).
Specific blocks of code reside on the industrial 0 It implements error diagnosis and handling.

controller in order to perform certain functions. These 0 It provides the user with the ability to issue

functions may be divided into three major categories: the commands to the network asynchronously via P user
functions that sampled values from the sensors of the node, friendly man-machine interface (MMI). Such
the functions that drive the actuators of the node and the commands include the dynamic change of the
functions that perform certain processing on specific parameters of the local tasks residing on nodes and
values. A sequence of functions called one after another upload and download facilities that allow monitoring
forms a task. Tasks also reside on nodes. and debugging of the application.
The tasks that exist on the node may be self-sufficient, An interactive and user friendly tool has been built
which means that all their inputs are produced on the node, enabling the elaboration and the evaluation of the
or they may need values produced in other nodes. In the application operating environment.
former case we have to deal with 'local memory', meaning These feames make the implemented system capable of
the memory area of the node, where the values of the handling industrial applications successfully even under
sensors and actuators of the node that are not needed to adverse conditions.
other nodes are stored, while in the latter case we have to Emphasis must be placed on the fact that the end user is
deal with 'global memory' as well, meaning the node provided with a graphical editor to design the specific
memory area where the values of the sensors and the application. Thus the system becomes user friendly as the
actuators of the node that are used by other nodes are only thing the user has to do is just "draw" the specific
stored. application. This drawing is as easy as when drawing an
Finally and in order for the system to be more flexible, electronic design. The end user has simply to draw
tasks are parametric meaning that such information as for functional blocks and interconnect them in such a way that
example the sampling rate of a sampling function is a the specific logic of the application is met. The design
parameter of the system. Function parameters are stored in should give a complete understanding of both the functions
the so called 'parameter area' of the node. used and the flow of data. In order to be able to do this the
end user should always bear in mind the way the application
is structured, that is to say the structure based on tasks
111. THE NETWORK
MANAGEMENT residing on nodes, consisting of functions and using the
memory area of the node as the area to store their
parameters and their inputs and outputs either 'global' or
A factor indicative of the sophistication of an industrial 'local'.
application is its relative ability to implement complex The drawing of the application is an off line procedure.
functions that involve processing of values produced in The graphical editor creates all the necessary structures that
various spatially distributed nodes, taking into account time can be dubbed as the backbone of the network management
constrains. The more complex the functions required, the of the system. It also creates the necessary 8344 assembly
more sophisticated the network management should be, as it language code that has to be downloaded to the nodes of the
provides the backbone for the transfer of data via the system and that represent the specific tasks that must reside
network. on the nodes and whose execution is fundamental for the
The main features of the network management that has system.
been built are: The software structures created by the graphical editor
are continually updated by the network management

533

~~
program running as a background process. The main goal importance of these four kinds of commands certain
of the network management program is the polling of the priorities should be set for their service. That emphasises
'gbbal memory areas' of the nodes and the transmission of the fact that for instance the commands related to error
these data to all nodes that this is necessary. This polling is handling procedures are the most urgent ones and must be
done periodically and is dependent on the rate the 'global' executed as soon as they are issued. 'Ihe severest priority
data are produced. The transmission of data is done taking thus is reserved for error handling commands. Then
into account user requirements. The transmission rates must commands related to periodic reading and writing of the
not be greater than the polling periods of the 'global 'global memory areas' of the nodes are serviced. User
memory areas' and are detenmined by the 'time window' specific commands are of the third severest priority while
during which the user regards that the data must be commands about statistics are commands with the lowest
tKanSmitted. priority.
A tool is supplied to the user which allows the The mechanism that has been implemented in order to
monitoring of the functioning of the application. This tool resolve this priority scheme is one based 011 four software
monitors the situation of the network and gives the user the FIFO buffers. The FIFO buffer mechanism allows dynamic
capability to watch the various parameters that are change of the priority scheme as well, in the case of the
associated with the network functioning at run time. It also application requiring to do so. This could for instance
enables the dynamic change of specific elements like the happen when the periodic reading and writing commands
parameters of the tasks. This tool may serve as a debugger consume a percentage of the CPU time that prohibits the
as well, since it allows the user to issue specific commands execution of user specific commands. In that case a
at run time to check certain properties of the network temporal change in the priority scheme solves the problem.
functioning. The data that the BITBUS manager continually collects
In the case of an error occurring in the network during are stored in the global heap area of the master node. From
the execution of a command, a response containing an error there on, they can be very easily accessed by any
code that completely specifies the type of the error is application that can behave as a Dynamic Data Exchange
returned by the network. If the error is a recoverable one, (DDE) client. That is so because the BITBUS management
then the BlTBUS manager program broadcasts all necessary program is implemented in such a way as to behave like a
commands to the network so that normal functioning is DDE server. This means that it can be hot linked to any
restored. If the error cannot be recovered then a message is application that is a DDE client, or in other words whenever
shown on the central station screen informing about the the 'global memory areas' data values that are stored in the
nature of the error and the actions required by the user. An global heap change, all client applications linked with the
example of the former type of error is the situation when a BITBUS management server are automatically informed
node is not responding properly. Then the network manager about the changes and are passed the new data values. This
program will inform the user about the situation faced and it DDE mechanism allows transmission of only the data that
will try to reset the node. If its attempts are successful then have changed surpassing this way the useless and time
normal functioning of the network resumes automatically. devouring procedure of transferring a large number of
If the network manager does not succeed in solving the unchanged data. It must be noted that the client applications
problem encountered, which means that a hardware problem may be anything from a database to a visualisation program
of some kind has taken place (for instance in the advent of and one can easily understand how important this way of
physical link damage) the user is informed about the organisation of the BITBUS management program is.
network problem and the actions required. Sparingly no limitations at all, as of what kind of processing
Furtheamore the network manager program is of the 'global' data values may take place or as of how these
responsible for the collection of statistical data about the values may be presented to the end user, exist.
functioning of the network. Such data include the number None of the above could have been accomplished of
of data pachts transmitted and received, the number of course if it had not been for the usage of the Microsoft
errors occurring and times associated with the transmission Windows, as the environment under which our application
and reception of data. These data can be presented was developed. Such Microsoft Windows aspects as the
graphically at run time giving a way to evaluate the network multitasking capabilities and the Dynamic Data Exchange
performance. provide our system with dynamic tools and made possible
Clearly there are four kinds of commands that may be the integration that has finally been accomplished [lll. A
issued to the network. The first kind is about the regular factor that has largely contributed to the final result is the
polling and writing of the 'global memory areas' of the principles of the object oriented programming that the
nodes. The second is about specific user commands. The whole program was based on. This way of building the
third is associated with the error handling management. program made it more structured and gave it greater
Finally the fourth kind is related to the acquisition of capabilities and efficiency than conventional ways of
statistical data. Because of the difference in the relative programming.

534
 10. Provides the user with an easy, universal way of
acquiring the data needed as it is based on the Dynamic
IV.CONCLUSIONS Data Exchange principles of the Microsoft Windows
environment.
The presented system has been installed and is working
In this paper a system based on the industrial local area successfully in five big Greek industries contributing to the
network Intel BITBUS is presented. The system that has modernisation of the Greek industry.
been implemented :
1. Provides high reliability as it guarantees that the
periodic exchange of data through the network will not REFERENCES
exceed given time intervals.
2. Reduces cabling costs as well as design, installation [l]. IXe MAP Book An Introduction To lk'wtd
' Networking,
and adjustment costs while making maintenance and Inustrial Networking IncorpoIlrted,1987
modification easier. [2]. H A Schutz, "'he Role of MAP in Factory Integration",ZEE
3. Permits advanced and remote control. Tnmsactionr on brdurtrial Electronics Mag., Vol 35, No. 1, pp. 6-13,
February 1988.
4. Is based on the idea of managing a spatially [3]. FIP TechnicalDesc+ion, FIP Club.
distributed data base with respects .to specific time [4]. C zebennann and H. Rake, PROFIBUS, German Stmrdanl In
constraints. Open &-Site Communication
5. Provides powerful, intelligent nodes. (51. The BITBUS hfzrconncct Serial Control Bur Specification,
Intel colporation1988
6. Provides the capability of error diagnosis and [6]. J. Henshall and S . S a w , OSI Explained: End-To-End Coquter
correction. CommunicationStat&&, Qlis Horwood Limited,l988.
7. Presents the user with a friendly, easy to use, [7]. ISOnEC 9505M-factwing Message Speczfiation
graphical way to draw the specific application. [8]. Ec STRIDELIGHTProposal 1992.
[9]. Microeomputerproducrs, Intel corporation,1992.
8. Is dynamic as it allows the change of the parameters [101. iDcx 51 Distributed Control fiecuthe User's GuidepReleaw
of the tasks residing on the network nodes at run time. 2.0, Intel Corporation,l987.
9. Is an open system as it does not put any specific [ll]. Peier Norton and Paul Yao, B o r W C++ Programming for
limitations during the design of the application. Windows, Borland Bantam, 1992.

535

~ ~