Fault Tolerance Identification of Application Layer Protocols Based on Protocol Natural Redundancy Decoding

Aiming at the problem that the existing network traffic classification technology mainly relies on the data of low or even no error packets, this paper proposes a fault-tolerant identification method for application layer protocols based on natural redundancy decoding under high error conditions. Firstly, the potential redundant information in the protocol layer that can be used for auxiliary decoding is mined, and then the natural redundant decoding model is established. On this basis, the maximum a posteriori probability criterion is used to derive the optimal decoding formula, and the corresponding fast decoding algorithm is designed. The test results on the UNSW-NB15 dataset show that the proposed method can successfully identify the protocol type under the condition of 10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> bit error rate, which provides a new idea and means for robust traffic classification in complex communication environments.