Research On Intelligent Parsing Of Unknown Protocol In Industrial Internet Of Things

Industrial digitization and intelligence have driven the Industrial Internet of Things to open and interconnect,and the network topology has become increasingly complex.The transmission requirements of industrial equipment have changed accordingly,causing many unknown protocols,including new protocols and proprietary protocols,to emerge in the Industrial Internet of Things.The format specifications of unknown protocols are usually unknown,and traditional protocol parsing methods cannot be used to parse,which brings inconvenience to data transmission between equipment and security detection of industrial systems.Therefore,the parsing of unknown protocols has become an urgent research problem in the industrial Internet of Things field.This thesis studies the intelligent parsing mechanism of unknown protocols for the Industrial Internet of Things from two aspects: protocol identification and protocol message parsing.For protocol identification,this thesis proposes an unknown protocol identification scheme based on online broad learning.It uses text feature extraction technology to mine the message features of industrial protocols,builds a protocol classifier based on broad learning,and use the improved similarity measurement scheme to compare the protocol representative samples corresponding to the classification result with the unknown protocol message to verify the classification result and feed it back to the classifier to realize the evolution and update of the model.For protocol message parsing,this paper proposes an unknown protocol message parsing scheme based on text pattern matching,using clustering center extraction algorithm and valued field inference algorithm to mine the different format specifications and corresponding representatives that may exist in the same industrial protocol,constructing protocol field templates.It uses the text-similarity measurement algorithm to match the templates during online parsing,and parse the protocol messages according to the format specifications of the templates.The solution proposed in this thesis could parse protocol without prior knowledge,and the model used has self-learning capabilities,which is suitable for the parsing of unknown protocols.The scheme is compared with several classic algorithms and schemes through experiments.The experimental results show that this scheme can accurately extract the message characteristics of industrial protocols,accurately match the templates,and achieve higher-precision intelligent analysis of unknown industrial Internet of Things protocols with lower delay.