An experiment in the design of a microprocessor-based distributed discrete simulation language.
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1986.
|
| Subjects: | |
| Online Access: | Link to ProQuest copy Link to OAKTrust copy |
| Abstract: | The research reported in this dissertation involved the design and development of a multitasked emulation of a microprocessor-based distributed discrete simulation system based upon the support function approach. A prototype system was developed using a version of SIMPAS. The SIMPAS preprocessor was modified to provide automatic decomposition of a sequential simulation program into a distributed one for execution on a TI 990/12. This system achieves the decomposition of SIMPAS simulation programs with relatively little effort from the programmer and is a useful and practical approach to developing and partitioning DDS/SF applications. The support function approach to distributed discrete simulation (DDS/SF) is characterized by the partitioning of simulation support activities to the nodes of a distributed system. In the prototype system, simulation support activities consisting of queue management, statistics collection, and random number generation and the model-specific event routines were grouped as separate processes during implementation. To allow observation of the prototype system without complications resulting from the distributed hardware, the prototype was implemented in a multitasked environment using the same functional decomposition as would have been required on the target microprocessor network. To determine the impact of the chosen partitioning upon the prototype DDS/SF system, sample simulation applications were executed and the communication generated was observed and recorded. These communication measurements were then analyzed to determine the characteristics of communication traffic. The analysis revealed that patterns exist in the communications generated by a distributed simulation program based upon the type of model executed, its length, and its traffic intensity. Communications were also examined to determine the impact of their urgency upon the performance of the system. Results show that an average of almost 70% of the simulation support activities of the tested applications implemented were requests for anticipatory or non-interactive activities. Through the lessons learned from this implementation and examination of the completed prototype system, the requirements of a preprocessor-based DDS/SF system were examined to determine if changes in the design or implementation strategies might be beneficial. These design alternatives include changes in partitioning, in the communication protocols, and in the implementation language. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. |
| Physical Description: | xiii, 161 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references (leaves 112-118). |