Research And Implementation Of Deep Packet Filtering Technology

It's a problem of controllability in network information security that the content of information should be healthy and harmless to people. To filter the content of the information when it being spreaded in network is an important approach which could assure the network information spread in a healthy and regulated way.In order to bypass the traditional exact string matching based filtering product, the releasers of junk information and illegal information might transform or disguise their malicious information before releasing it into the network. So it is necessary to employ some effective technique which can recognize those transformed or disguised malicious information in an intellectualized way. But it is a key technique and a bottleneck of performance in content filtering to match multi-pattern approximate strings with the network information now. In addition, with the rapid development of network, a network which has high bandwidth demands its packet filters have high performance.In this paper, after studed some common string matching algorithms and many matching related hardware structures, a hardware oriented multi-pattern exact string matching algorithm has been presented. The main ideal of this algorithm is to take advantage of the parallel work trait of hardware to match each pattern character with the input character simultaneously, and then to produce their matching result iteratively based on the information produced in previous turns. Based on this exact string matching algorithm and based on the study of the coding regularity of the characters in computer, an approximate string matching algorithm that can filter out interferential characters has also been presented.Applied the two algorithms, a matching chain structure has been designed, used this matching chain structure, a exact matching based search engine and a search engine which can filter out interferential characters have been designed with FPGA. The simulated result of the search engines indicated that the matching chain circuit is simple and could be implemented easily, and it can match the text string with multi arbitrary length patterns parallelly and effectively.To satisfy the needs of built a practicable content filter, associated with pre-transform the characters, a packet filter base on TCAM has also designed, which can implement the...