臺灣博碩士論文加值系統

早期模擬系統獨立在電腦上進行模擬，隨著模擬應用單元的開發及系統的複雜化，獨立電腦的執行成效會呈曲線下滑。解決執行成效不彰的方式就是將傳統獨立於個別電腦的模擬執行，開發成為可以藉由網路連結，分散於不同主機執行的模擬。分散式模擬藉由網路互相傳遞訊息而達到資料的交換及互動，分享在時空上一致的虛擬環境，但卻因此必須付出資料於網路上的傳輸與時間上同步的代價。資料分佈管理的功能就是減少無關資訊的傳輸，是降低分散式模擬成本的方法之ㄧ。本研究希望能夠減少分散式聯網模擬付出的資料分佈管理成本，提升分散式聯網模擬的執行效能。
資料分佈管理機制中的核心是資料區域的重疊比對，本論文聚焦在網格法分割路由空間成為矩形細胞以區域的重疊比對作為研究基礎。結合訊息於網路的傳輸時間與計算區域的重疊比對時間，找出網路與區域重疊比對成本最佳化的細胞數量組合。網格法的更新欄位成本隨著細胞數量的增加而增加，影響更新欄位成本的因素在於資料的區域範圍與細胞重疊數量的多寡，當資料區域範圍較大，則需記錄資料的細胞數量增加，因此更新欄位成本就會隨著區域範圍增加相對地增加，本論文將以網格法對於資料區域的重疊比對方法，控制影響總成本的因素，執行模擬獲得成本最佳化的組合。

The execution of simulation is implemented in a single computer. However, the performance of simulation in a host will diminish when the simulation units increase and the simulation system complexity raise. The resolution is to develop distributed simulation achieved by networked messages passing to exchange data and interact each other. Nevertheless, distributed simulation sometimes suffer the loads of data communication and synchornization control in order for maintaining the consistent simulation environment. This work focus on data distribution management for the reduction of communication burden by means of shrinking the irrelevant data passing to increase the distributed simulation performance.
The core of the data distribution management mechanism is the region- overlapping computation. This thesis exploit the grid-based method which divide the routing space into equal cell to compute the overlap regions. Considering the factors of region-overlapping computation time and data transmission time, this paper presents an analysis of optimal cost with the number of cells. The cost of region-overlapping computation go up when the cell number increase because the field-updating cost in cells rise. But, the irrelevant data in communication networks decrease. This thesis analyze the cost of grid-based data distribution management in terms of cells number, the size of update and subscribe region, and the simulation unit number. Finally, we simulate the grid-based data distribution management environment using different factors to find an optimal cost.

目錄
誌謝 ii
摘要 iii
ABSTRACT iv
目錄 v
表目錄 vii
圖目錄 viii
1. 緒論 1
1.1 研究動機與背景 1
1.2 聯網模擬發展近程 1
1.3研究目的 2
1.4 研究方法 4
1.5 論文組織 4
2. 文獻探討 5
2.1分散互動模擬簡介 5
2.2 高階架構 8
2.3 資料分佈管理 12
2.3.1 區域法 17
2.3.2 網格法 20
2.3.3 複合法 25
2.3.4 三種方法之優、缺點 28
3. 資料分佈管理成本最佳化 30
3.1 細胞的儲存機制 33
3.2分散式資料分佈的環境與架構 35
3.3 模擬想定 41
3.4 演算法詳述 41
4. 模擬實驗結果及分析 48
4.1 實驗設備 48
4.2 模擬環境設定 48
4.3 實驗結果 51
4.3.1 細胞數量增加對總成本影響 51
4.3.2 更新區域範圍增加對總成本的影響 54
4.3.3 訂閱區域範圍增加對總成本的影響 55
4.3.4 坦克物件數量增加對總成本的影響 56
4.4 實驗結果之分析 58
5. 結論與展望 61
參考文獻 62
自傳 64

參考文獻
[1] “IEEE Standard for Modeling and Simulation High Level Architecture (HLA)- Framework and Rules,”IEEE Std. 1516-2000, Sep., 2000.
[2] “IEEE Standard for Modeling and Simulation High Level Architecture (HLA)- Federate Interface Specification,”IEEE Std 1516.1-2000, 2001.
[3] “IEEE Standard for Modeling and Simulation High Level Architecture (HLA)- Object Model Template Specification,”IEEE Std 1516.2-2000, 2001.
[4] http://hla.dmso.mil/。
[5] “IEEE Standard for Information Technology - Protocols for Distributed Interactive Simulations Applications,”IEEE Std 1278-1993, 1993.
[6] http://www.dmso.mil/dmso/docslib/mspolicy/msmp/。
[7] 傅光聖，“在網路虛擬實境中對物體狀態的分散式互動模擬之改良”，碩士論文，國立台灣大學資訊工程研究所，2000。
[8] Cohen, D., and Kemkes, A., “Applying User-Level Measurements to RTI 1.3 Release 2, ” Proceedings of the Simulation Interoperability Workshop, 1998.
[9] Tan, G., Xu, L., Moradi, F., and Zhang, Y., “An Agent-Based DDM Filtering Mechanism”, in Eighth International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems , 2000.
[10] Berrached, A., Beheshti, M., and Sirisaengtaksin, O., “Evaluation of Grid-based Data Distribution in the HLA,” Proceedings of the 1998 Conference on Simulation Methods and Applications, Orlando FL, pp. 209-215, 1998.
[11] Tan, G., Ayani, R., Zhang, Y.S., and Moradi, F., “Grid-Based Data Management in Distributed Simulation,”Proceedings of Thirty-Third Annual Simulation Symposium, Washington, U.S.A., pp.7-13, 2000.
[12] Boukerche, A., Roy, A., and Thomas, N., “Dynamic Grid-Based Multicast Group Assignment in Data Distribution Management, ”in Fourth IEEE Distributed Simulation and Real Time Applications, DS-RT, pp.47-54.
[13] Boukerche, A., Roy, A., “Dynamic Grid-Based Approach to Data Distribution Management,” Journal of Paralled and Distributed Computing, pp.366-392, 2002.
[14] Rak, S. J., and Van Hook, D. J., “Evaluation of Grid-Based Relevance Filtering for Multicast Group Assignment,” Proceedings of the Distributed Interactive Simulation, 1996.
[15] Tan, G., Ayani, R., and Moradi, F., “Optimizing Cell-Size in Grid-Based DDM”
Proceedings of the Fourteenth Workshop on Parallel and Distributed Simulation, 2000.
[16] Lu, T., Lee, C., Hsia, W., and Lin, M., “Supporting Large-Scale Distributed Simulation Using HLA,” ACM Transactions on Molding and Computer Simulation , 2000.
[17] Tan, G., Zhang, Y., and Ayani, R., “A Hybrid Approach to Data Distribution Management,”in IEEE Distributed Simulation and Real Time Applications , 2000.