行動代理者（Mobile Agent）是可在一異質網路系統（Heterogeneous Network System）內從一主機（Host）遷移（Migrate）至另一主機並和其他代理者及分散資源系統（Distributed Resources）溝通及交互作用（Interact）之軟體程式。它們特別適合用來發展分散式及網路上的應用系統。
　　行動代理者雖已有許多相關研究，但由於網路本身就是不可靠的，而且不管是行動代理者本身或是其執行平台（Platform）都有可能遭到惡意的程式破壞，使其產生不正確的結果，甚至損毀整個系統。這一些因所導致的果，皆可歸類為拜占庭錯誤（Byzantine Faults）。拜占庭錯誤在這裡指的不管是系統平台的軟、硬體失敗或經由惡意攻擊造成執行平台錯誤所展現出的種種行為，例如送出偽造的行動代理者，或將代理者的運算執行錯誤，導致產生非預期正確的結果…等。拜占庭容錯（Byzantine Fault Tolerance）是以副本（Replica）的方式（假設系統中有n個副本）容許系統中不超過└(n-1)/3┘個拜占庭錯誤時，保證系統結果的正確性。因此將拜占庭容錯應用至行動代理者系統上，便可大大的增加系統可靠性。
　　本研究最主要的目的是（一）將拜占庭容錯演算法和現有的行動代理者系統做整合，並改善演算法中群組管理方面的問題，使得群組狀態可以即時更新，增加演算法的穩定和可靠性；（二）在MACE（Mobile Agent Carrier Environment）中加入群組概念，達到行動代理者傳送時的通透性（Transparent）；（三）在MACE中加入回復機制（Recovery Mechanism），復原已由拜占庭協議（Byzantine Agreement）確認（Commit）過的行動代理者。（四）最後再針對所加入的拜占庭協議模組（Byzantine Agreement Module）做一行動代理者系統的整體效能評估。

　　Mobile agents are software programs that can move from host to host and interact with other agents and distributed resources in a heterogeneous network. They are therefore particularly effective for developing distributed applications.

　　Although there are already a lot of researches on the reliability of mobile agents themselves, the network, i.e. the transmission medium, itself is by no means reliable. Mobile agents and their executing platforms may be attacked by malicious programs, making them to produce incorrect results, and to the extreme, damaging the whole system. These faulty behaviors can all be classified and modeled as Byzantine Faults. A Byzantine-fault-tolerant system with a total of n replicas can tolerate up to └(n-1)/3┘ simultaneously faults. So it would be a good idea to incorporate Byzantine-fault-tolerance on a mobile agent system to increase its reliability.

　　The purpose of this thesis is 1) to integrate the Byzantine-fault-tolerance algorithm with an existing mobile agent system – the MACE (Mobile Agent Carrier Environment) – and to improve its group management capability, 2) to add group mechanism to MACE in order to achieve transmission transparence of mobile agents, 3) to add a recovery mechanism to MACE for recovering the system from any Byzantine Fault, and 4) to perform the evaluations on the entire system with the newly implemented Byzantine Agreement Module.

[1] Miguel Castro and Barbara Liskov, “Practical Byzantine Fault Tolerance”, Proceedings of the Third Symposium on Operating Systems Design and Implementation, New Orleans, USA, February 1999

[2] Miguel Castro and Barbara Liskov, “Authenticated Byzantine Fault Tolerance Without Public-Key Cryptography”, MIT Technical Memo MIT-LCS-TM-589, Massachusetts Institute of Technology, June 1999

[3] Miguel Castro and Barbara Liskov, “A Correctness Proof for a Practical Byzantine-Fault-Tolerant Replication Algorithm”, MIT Technical Memo MIT-LCS-TM-590, Massachusetts Institute of Technology, June 1999

[4] Miguel Castro and Barbara Liskov, “Proactive Recovery in a Byzantine-Fault-Tolerant System”, In Fourth Symposium on Operating Systems Design and Implementation (OSDI), San Diego, USA, October 2000

[5] Miguel Castro, “Practical Byzantine Fault Tolerance”, Ph.D. thesis, Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, January 2001

[6] Leslie Lamport, Robert Shostak, and Marshall Pease, “The Byzantine Generals Problem”, ACM Transactions on Programming Languages and System, pp.382-401, July 1982

[7] Nancy A. Lynch, “Distributed Algorithms”, Morgan Kaufmann Publishers, Inc.

[8] Taesoon Park, Ilsoo Byun, Hyunjoo Kim, and Yeom, “The performance of checkpointing and replication schemes for fault tolerant mobile agent systems, ”Proc. Reliable Distributed Systems, pp.256-261, Oct. 2002

[9] Pleisch, S. and Schiper, A., “Modeling fault-tolerant mobile agent execution as a sequence of agreement problems,”Proc. Reliable Distributed Systems, pp.11-20, Oct. 2000

[10] Pleisch, S. and Schiper, A., “FATOMAS-a fault-tolerant mobile agent system based on the agent-dependent approach,”Proc. Dependable Systems and Networks, pp.215-224, July 2001

[11] Pleisch, S. and Schiper, A., “Fault-tolerant mobile agent execution,”Proc. Computers IEEE Transactions on , pp.209-222, Feb 2003

[12] Rothermel, K. and Strasser, M., “A fault-tolerant protocol for providing the exactly-once property of mobile agents,”Proc. Reliable Distributed Systems, pp.100-108, Oct. 1998

[13] Wang, T. I. “A Mobile Agent Carrier Environment with Mobile Computing Facilities”, IIP: International Conference on Intelligent Information Processing, The 16th IFIP World Computer Congress, 21~25/08, 2000, Beijing.

[14] Wang, T. I. “A Mobile Agent Carrier Environment for Mobile Information Retrieval”, 11-th International Conference on Database and Expert Systems Applications - DEXA 2000, 05~08/09, 2000, Greenwich, London.

[15] Wang, T. I. “A Mobile Agent Carrier Environment”, ICS2000, 6-8 December, 2000, Chiayi, Taiwan, R.O.C.

[16] Wang, T. I. “以行動代理者支援行動資訊擷取”, 第五屆人工智慧與應用研討, Taipei, 17/Nov. 2000

[17] Jian Yin, Jean-Philippe Martin, Arun Venkataramani, Lorenzo Alvisi, Mike Dahlin, “Separating Agreement from Execution for Byzantine Fault Tolerant Services”, SOSP’03, Bolton Landing, New York, USA, October 2003

[18] Meng Yu, Peng Liu, and Wanyu Zang, “Intrusion Masking for Distributed Atomic Operations”, In Proceedings of the 18th IFIP International Information Security Conference (SEC’03), pp.229-240, May 2003

[19] 周晉立, ”Mobile Agent Carrier Environment”，國立成功大學工程科學研究所碩士論文，June 1998

[20] 徐永憲, ”Fault Tolerant Mechanism for Mobile Agent Systems”，國立成功大學工程科學研究所碩士論文，June 2003