طراحي يك سيستم تشخيص نفوذ به كمك روش هاي مبتني بر سيستمهاي هوشمند

Design of an Intrusion Detection System Using Methods Based On Intelligent Systems

ﺑﺸﺮ در ﺣﺎل ﮔﺬران ﺑﺮهه اي از ﺗﺎرﯾﺦ ﺧﻮد اﺳﺖ ﮐﻪ ﺳﯿﺴﺘﻢهاي ﮐﺎﻣﭙﯿﻮﺗﺮي ﺑﻪ ﺑﺨش ﺟﺪاﯾﯽ ﻧﺎﭘﺬﯾﺮ ازﺣﯿﺎت وي ﺑﺪل ﺷﺪه اﺳﺖ. ﺳﯿﺴﺘﻢهايﮐﺎﻣﭙﯿﻮﺗﺮي و ﺑﻪﻃﻮر ﺧﺎصﺷﺒكه هاي ﮐﺎﻣﭙﯿﻮﺗﺮي در ﺑﺴﯿﺎري از ﺟﻨﺒﻪ هاي زﻧﺪگي اﻣﺮوزي ﻧﻘﺶﺗﻌﯿﯿﻦﮐﻨﻨﺪه اي داﺷﺘﻪ و ﺑﻪ همين دﻟﯿﻞ اﻣﻨﯿﺖ آن ها ﺑﺴﯿﺎر ﻣﻮرد ﺗﻮﺟﻪ اﺳﺖ. ﺑﺮاي ﭘﻮﺷﺶ اﻣﻨﯿﺖ آن ها و در اﺻﻄﻼح ﺗﺸﺨﯿﺺ ﻧﻔﻮذ در شبكه هاي ﮐﺎﻣﭙﯿﻮﺗﺮي روش هاي زﯾﺎدي از ﺟﻤﻠﻪ ﺷبكه هاي ﻋﺼﺒﯽ ﻣﺼﻨﻮعي ﭘﯿﺸﻨهادﺷﺪه اﺳﺖ، اﻣﺎ يكي از ﭼﺎﻟﺶ هاي ﺷﺒكه هاي ﻋﺼﺒﯽ اﻓﺘﺎدن در دام بهينه هاي ﻣﺤلي اﺳﺖ. از دﻻﯾﻞ ﺑﺮوز اﯾﻦ ﻣﺸكل اﺳﺘﻔﺎده ازروش هاي مبتني ﺑﺮﮐﺎهش ﺷﯿﺐ در ﻓﺮآﯾﻨﺪ آﻣﻮزش اﺳﺖ ﮐﻪ در ﻧﺘﯿﺠﻪ ﺑﺮاي ﺣﻞ اﯾﻦ ﭼﺎﻟﺶ از روش هاي بهينهﺳﺎزي ﺑﺴﯿﺎري اﺳﺘﻔﺎده ميﺷﻮد. در اﯾﻦ ﻣﻘﺎﻟﻪ سعي ﺷﺪه اﺳﺖ ﺑﺎ اﺳﺘﻔﺎده از اﻟگورﯾﺘﻢ رﻗﺎﺑﺖ اﺳﺘﻌﻤﺎري ﯾك ﺷﺒكه ﻋﺼﺒﯽ ﭘﺮﺳﭙﺘﺮون ﭼﻨﺪ ﻻﯾﻪ را آﻣﻮزش داده و ﺑﻪ ﻣﻨﻈﻮر ﺑﺮرسي ﮐﺎراﯾﯽ آن، روش ﭘﯿﺸﻨهادي ﺑﺎ دو روش ديگر ﻣﻘﺎﯾﺴﻪ ﺷﺪه اﺳﺖ. ﻧﺘﺎﯾﺞ روش ﭘﯿﺸﻨهادي در ﻣﻮاردي ﮐﻪ ﺗﻌﺪاد ﻧﻤﻮﻧﻪها ﺑﯿﺸﺘﺮ ﺑﻮده اﺳﺖ بهبود 10 تا 15 درﺻﺪي را ﺑﻪ دﻧﺒﺎل داﺷﺘﻪ اﺳﺖ.

Humans are going through a period of their history that computer systems have become an integral part of his life. Computer systems, particularly computer networks, have played a crucial role in many aspects of today's life, so their security is highly regarded. In order to cover their security and the so - called penetration detection in computer networks, many methods such as artificial neural networks are proposed, but one of the challenges of neural networks is to fall into local. The main reason for this problem is the use of slope - based methods in the training process, which is therefore used to solve this challenge by many optimization methods. In this paper, we have tried to train a multilayer perceptron neural network using the colonial competition algorithm and in order to evaluate its efficiency, the proposed method is compared with two other methods. The results of the proposed method have improved 10 - 15 % improvement in cases where the number of samples was higher.