To solve the problems in traditional network protocol recognition methods, such as difficulty of manual feature extraction and low recognition accuracy, an application protocol recognition method based on Convolutional Neural Network (CNN) was proposed. Firstly, the raw network data was divided according to Transmission Control Protocol (TCP) connection or User Datagram Protocol (UDP) interaction, and the network flow was extracted. Secondly, the network flow was converted into a two-dimensional matrix through data prepocessing to facilitate the CNN analysis. Then, a CNN model was trained using the training set to extract protocol features automatically. Finally, the trained CNN model was used to recognize the application network protocols. The experimental results show that, the overall recognition accuracy of the proposed method is about 99.70%, which can effectively recognize the application protocols.

In order to extend the practicality of Augmented Reality (AR), a method based on Temporal Psycho Visual Modulation (TPSM) technology and digital camera to realize AR effect was proposed. First, the AR tags were embedded in the digital screen of the media. Based on the principle difference between the human eye to identify the perception and the digital camera to capture the image formed in the digital screen or projector, the digital camera equipment was used to obtain the digital screen image with AR tags which are not easily to be detected by human eye. Finally, the AR effect was displayed on the smart device that gets the AR tags. Simulation results show that the combination of AR and TPVM technology can accurately identify the AR tags in the image and achieve AR effect, while the human eye can not detect the AR tags. Through the mobile phone instead of 3D glasses and other extra equipment, the use restrictions of AR are reduced, and the practicality of AR is also expanded.

To solve the problem that the methods currently used to construct Chinese knowledge graph system are time-consuming, have low accuracy and require a lot of manual intervention, an integrated end-to-end automatically constructed solution based on rich data from Chinese encyclopedia was proposed, and a user-oriented Chinese knowledge graph was implemented. In this solution, some property and related text information of the original encyclopedia data were scraped to local system uninterruptedly by the custom Web crawler, and saved as a triple with extended attributes. Through graph-oriented database Cayley and document-oriented database MongoDB, the data in the archived triple files was imported in the back-end system, and then converted to a huge knowledge graph system in order to provide various services dependent on the Chinese knowledge graph in the front-end system. Compared with other knowledge graph systems, the proposed system significantly reduces the construction time; moreover, the number of entities and relations is at least 50% higher than that of the other knowledge graph systems such as YAGO, HowNet and the Chinese Concept Dictionary.

Cognitive network can improve the end-to-end performance of the network, and ensure QoS(Quality of Service) requirements. The existing routing algorithm does not have cognitive ability. To solve this problem, a multi-path routing algorithm of cognitiveload balancing was proposed, which combined the advantages of Q-learning algorithm and ant algorithm, to establish and maintain the route through ant algorithm, and to achieve congestion avoidance and load balancing by Q-learning algorithm. The simulation contrast with OPNET shows that the algorithm is valid and effective at controlling packet loss ratio, delay and bandwidth utilization.