Internet-of-Things application flow detection method based on deep learning

Abstract

The invention discloses an Internet-of-Things application flow detection method based on deep learning. The method comprises the following steps: carrying out structured data storage on a flow data packet on Internet-of-Things equipment through Wireshark; cutting and screening Pcap traffic packets by using a Split Cap tool, obtaining different cut samples of the traffic packets, converting traffic data with different fragment sizes into binary graph files for representation, and obtaining relatively accurate application traffic identification through adoption of a deep learning method. The method mainly comprises the steps of flow packet data screening, code cutting of a Pcap file, graph conversion of a structured flow data packet and application flow identification of a convolutional neural network. According to the method and the device, the application data packet in the Internet-of-Things equipment can be automatically captured and identified, powerful support is provided for flowtracking and tracing of the Internet-of-Things equipment, and the safety performance of the Internet-of-Things equipment is greatly improved.

Description

technical field [0001] The invention belongs to the field of computers, focuses on network security traceability at the device side of the Internet of Things, and proposes a deep learning-based Internet of Things (Internet Of Things, IOT) application flow detection method, which can detect objects completely and accurately according to the information contained in the flow Methods for networking application traffic types and sources. Background technique [0002] Traffic inspection is the work of associating network traffic with applications, and is an important task in the field of network security. In the field of network security, traffic classification is actually the initial step in anomaly detection such as querying malicious network resource usage, and is an essential part of IoT security detection. Currently, there are four main traffic classification methods: port-based traffic analysis, Deep Packet Inception (DPI)-based traffic analysis, statistics-based applicati...

AI Technical Summary

Problems solved by technology

Although the classic machine learning method solves many problems that cannot be solved by the rule-based method, such as encrypted traffic classification and high computing cost, it faces the challenge of a large number of traffic similar features, which will lead to inaccurate traffic classification.

[0003] Due to the complex security behavior of IoT devices, it will be difficult to monitor and identify traffic using traditional methods

However, IoT devices rarely have professional security traceability methods, and it is difficult to ensure accurate identification of application traffic. Only some more traditional rule matching methods can be used. However, due to the popularity of encryption algorithms at this stage, the recognition accuracy of rule matching methods is often not as good as Satisfactory, there is an urgent need for high-precision detection methods to detect application traffic at this stage

Images