在不可靠網路中應用分散式演算法解決火力分配決策問題之研究

Abstract

在戰場上，所有的決策都必須非常迅速，為了能在短時間找到最佳決策， 通常都會在火協機構放置cluster加速運算，減少計算的時間。但是當所有電腦統一放在同一個地方，容易遭受到攻擊，為了要避免此種情形，我們將各個運算的電腦，分散到各個火力單位，形成分散式架構，來解決容易遭受攻擊的問題。但當火力單位在移動或通訊設備遭到破壞時，導致各電腦的通訊不穩定，分散式架構在不可靠的環境下，會大大影響了計算的效能，與決策的統一性。在本論文中，我們利用Peer-to-Peer的分散式架構，並結合啟發式演算法來解決不可靠環境下所產生的問題。另外，我們以戰場上常見的火力分配決策問題 ，來驗證我們的系統是否可靠。

All the decisions on the battlefield must be very fast. To find the best decision in a short time, usually place cluster architectures in the decision center to reduce the calculation time. However, when we put the computing resources in the same place, it will be attacked easily. In order to solve the problem of being attacked easily, we distribute computing units to various weapon to form a distributed system. But when weapons were moving or communications equipment was destroyed by enemies, the communication will be very unstable. The distributed system is not efficient on division of work and integration in the communication unreliable environment. In this thesis, we use distributed system of Peer-to-Peer Combined with heuristic algorithms to solve problems in the communication unreliable environment. In addition, we use weapon target assignment problem as an example to verify our system is reliable.