Implementation of Information Security Management System with Software Agents

[[abstract]]由於網路技術的快速發展，寬頻網路的應用也越來越普及，使得各種資訊安全管理的問題應運而生，像是電腦中毒不斷大量傳送封包而造成網路的壅塞，或是非經授權使用他人電腦進行資料竊取，又如安裝未經企業授權的應用軟體，甚至因為電腦安裝雙網卡而使得攻擊者不需透過防火牆入侵企業網路，這種種的模式對企業網路造成無比的危害，一般而言企業網路的安全多半著墨在防火牆上，其實真正最大的資安危害是發生在企業內部本身。 本篇論文，以實現自動化代理器之資訊安全管理系統為目標，透過代理器值入使用者電腦，進而蒐集使用者軟硬體資訊，自動判斷使用者是否符合該企業資訊安全政策，藉由不提供未符合企業資訊安全政策的員工網際網路服務，以加強員工安全使用電腦的習慣。在管理者功能上，可提供管理者查核使用者電腦軟硬體資訊及企業資訊安全政策的設定，以有效管理企業內部的資訊安全及電腦資源。 根據實際系統架設及實驗顯示，本論文所提出之自動化軟體代理器，可以加強使用者安裝修正程式及資訊安全政策設定的意願及速度，所以能減少企業內部電腦中毒而引發網路壅塞或被惡意值入ActiveX元件的事件減少，對於未安裝修正程式及未符合公司資訊安全政策的使用者名單，管理者可加以通知使用者進行修正，為了加強智慧財產的觀念及有效率的工作，並提供所有用戶端安裝軟體的統計報表。[[abstract]]As the internet advances , the applications of the wide-band network are widely spread. This results in different types of problems of information security management. For example , the network might be congested by a lot of re-transiming packages due to some virus-infected devices existing in an enterprise network. Unauthorized processes might also be inserted into the system to steal important and valuable information is some ways , even when the network is supposed to be protected by firewall software. For most enterprise network , the security relies on the firewall software , however , most problems of information security are happened inside enterprises. In this thesis, we propose a systematic method to develope an information security management system with the help of software intelligent agents , which are installed in advance on client-side computers to collect user's software and hardware installation information. An intelligent database server is also design to provide the security checking rules for those agents. Such that all the computers connected inside the network are forced to obey the security policies of enterprise. The software agents instantly correct the unsafe setting of client-side software and routinely report the status of security to the users. Accordingly the proposed system is supposed to keep herself safe on information security via the corporative distributed software agents. Experimental results demonstrate the effectiveness of the proposed method