Architecture

Training Simulator ofa

based on

Distributed Processing

Shi-Yu Gong, Liang-Cai Liao and Ya-Xin Han, National University of Defence Technology

ABSTRACT

The potential abilities of micro-computers make it possible that low cost training simulators are available. Distributed processing is an effective way for implementation of this kind of simulator. This article presents a training simulator architecture based on distributed processing and discusses its role in the training simulation support system.

INTRODUCTION

With digital control technology becoming increasingly popular, there is a new and higher requirement for human performance in operating and control processes. As an effective instrument for this purpose, the training simulator is being widely used. Its advantages include:

1) Modelling various complex conditions and having a definite purpose;

However, as far as developing a real training simulator is concerned, the development activities often are a complex, technical, expensive and time-consuming process, involving various knowledge and personnel. It is important to develop a training simulator in order to achieve the original intent. In this respect, Hays [l] gives us meaningful inspiration. Based on a technological perspective, this article discusses the question of whether a training simulator hardware sub-system can be made from commercial digital equipments (mainly micro-computers and local area network products) so as to noticeably lower costs. Distributed processing is an effective tool for implementation of this kind of simulator. According to popular thought, a training simulator architecture is suggested. Finally, some requirements about developing a training simulation support system based on the architecture are discussed.

DESIGN OF THE ARCHITECTURE 2) Low training cost; 3) Advance training; and 4) Efficient training, easily evaluated and improved. The development of a real training simulator is often

influenced by technical, as well as, financial factors. Sometimes, these applications are limited to micro-computer technology growth and the ratio of performance and price rise make it possible to solve the complex model by means of micro-computer systems.

Authors' Current Addresses National University of Defence Technology, Department of Systems Engineering, Changsha Hunan, P R China 410073

Based on a presentation at NAECON '96

0885-8985/96/ $5.00 0 1996 IEEE There exists the opportunity that the simulator, being suitable for training needs, can be developed at low cost.

8 IEEE AES Systems Magazine, September 1996

PERFORMANCE DEMANDS

node 1

The training simulator is a substitute for the actual system and serves an all-purpose, man-machine interactive instrument. The total training simulation system consists of software and hardware. In addition, the system includes personnel (instructor and trainee), training program, operating and control rules, etc. The essential demands for performance of the training simulator are as follows: 1) Real-time solving for the system model; 2) Real-time response to operating and

3) High-fidelity presentation of system status; and 4) Exact duplication of actual man-machine interaction.

Therefore, the simulator should be a multi-user and multi-task processing system. It should provide high-speed information processing, information acquisition and information transmission abilities. It must be rich in the functions of graphicdimages and voice processing.

control commands;

rivde 2 . . . node N

ARCHITECTURAL DESIGN

In light of the demanding performance and letter-perfect training function, an architecture of the training simulator based on micro-computer and local area

Fig. 1. Topology of the Architecture

networks is suggested; its topology being bus-type and its connections by cable as shown in Figure 1.

solver; instructor console; or trainee equipment. The trainee node may be equipped such specific items as a key-board, large-size screen, special controls, etc. The model solver node uses commercial micro-computers such as AST, COMPAQ, etc. The criteria for the micro-computer performance arc based on processing requirements, e.g., multi-medium functions. System software mainly includes Windows 3.X, Netware Lite, Borland C+ + .

System communication within the point-to-point manner; the data are transmitted at the rate of 10 MB/Sec. Each type of node has its own procedure for sending and receiving data. It is the system information related to the node that can be exchanged.

Each node can be defined as one of three types: model

Model Solver Node

model; varying system status; and modelling the actual system’s behaviour.

The model server node serves as a a problem-solving

Instructor Node

accepts instructor’s commands; activates training programs; directs training processes; estimates the effects of training.

The instructor node presents the system status;

Trainee Node

accordance with actual conditions; and implements operating and control interactions.

The training node presents the sub-system status in

DISTRIBUTED PROCESSING

Except for the instructor node, the number of other nodes could exceed more than one in the system. That is, the whole training simulation task may be distributed to each node and carried out cooperatively by them. This is owing to the architecture’s choice and make it effective to simulate more complex systems.

The system model is developed on the basis of object-oriented thought. Therefore, it is the model of an element, or some in the system, object set that is solved in the model solver node. In this way, the model solver node is related to the other nodes only by means of information exchange. With the task being carefully designed, each trainee node carries definite functions, such as, interaction between personnel and equipments, pre-processing the operating and control information, etc. Similarly, there exists information exchange between trainee nodes and other nodes.

CONSIDERATION OF THE SIMULATOR SUPPORT SYSTEM BASED ON THE ARCHITECTURE

If a training simulator of high quality and low cost is able to be developed quickly, it is necessary to provide for a training simulator support system. Besides the general functioning of this system [2], the development of a training simulation support system based on the architecture must include consideration of distributed processing. Its modelling methodology [3] should be based on object-oriented thought. For the moment, we suggest a development paradigm depicted (on next page) in Figure 2.

After the requirement for training simulation is submitted, the two libraries are observed. If the set of objects and/or sub-models in the libraries is enough to be organized into the required system model, the task of building the model ends. The models and objects are reusable. Otherwise, by means of object analysis, specification and generation. new objects can be joined into the object library until there is enough to build the system model.

be integrated in the training simulation support system. They could be simulating the control and management of the sub-system; instructor support sub-system; trainee

General supporting and processing functions should

IEEE AES Systems Magazine, September I996 9

SIMULATION TRAINING REQUIREMENT

1 1 1

OBJECT SPECIF'ICATICN

OBJECT GENERATION

OBJECT LIBRARY I c c

OBJECTS ORGANIZATION ' T MODEL LL4RARY

1 SYSTEM MODEL

1 SlhiLTLATION TEST

T W N I N G SIMJLATION

Fig. 2. System Model Development Paradigm

operating and control interfaces; communication protocol and supporting software, etc.

CONCLUSION

With the ratio of performances and prices of micro-computers as well as local area network products

rising steadily, the low-cost training simulator is being used to advance the simulation technology application. However, distributed processing is an effective way to model the complex system to carry out training simulation tasks. The architecture of the training simulator suggested in this article can supply a suitable environment. The training simulation support system based on the architecture should use object-oriented thought as its methodology base. Thus, a training simulator can be effectively developed.

REFERENCES

[l] Hays, R.T., 1992, "Systems Concepts for Training Systems Development," IEEE Transactions on Systems, Man, and Cybernetics, Vol. 22, No. 2, pp. 258-266.

[2] Balmer, D.W. and Paul, R. J., (1990), "Integrated Support Environments for Simulation Modelling," In Proceedings of the 1990 Winter Simulation Conference, 0. Balci, R.P. Sadowski and R.E. Nance, Eds., pp. 243-249.

[3] Nance, R.E. and Arthur J.D., (1988), "The Methodology Roles in the Realization of a Model Development Environment," In Proceedings of the 1988 Winter Simulation Conference, M. Abrams, P. Haigh and J. Comfort, Eds. pp. 220-225.

Shi-Yu Gong is a lecturer in the Department of Systems Engineering at the National University of Defence Technology in China. He received the B.A. degree in computer science from the University of Northeast, Shenyang, China in 1983. Seven years later, he received the M.A. degree in system engineering from the National University of Defence Technology in China. His interest is mainly devoted to complex system modeling, modeling methodologies, simulation, and simulation support systems.

Liang-Cai Liao is an assistant in the Department of Systems Engineering at the National University of Defence Technology in China. He received the B.A. and M.A. degrees in system engineering at the National University of Defence Technology, China, in 1991 and 1994, respectively. His main research interest is complex system modeling, simulation and simulation software systems.

Ya-Xin Han is a postgraduate student. She received the B.A. degree in information system engineering from the National University of Defence Technology, China, in 1994. She is returning to research in systems engineering.

IO IEEE AES Systems Magazine, September 1996