Analysis And Research Of Mobile Internet Users Data

With the rapid development of mobile Internet, the number of mobile Internet users and user data appears explosive growth, Research data of the mobile Internet can help to understand the operational status of the network, to understand user needs and potential demand, to infer network business trends, to achieve fine control operations. Telecom operators also proposed "users can identify applications can be distinguished, flow can be control, network can be manage" four goals. Some of the existing methods of data analysis is not mature in terms of massive data processing and real-time data stream processing, especially the accuracy, flexibility and scalability of applications identification.This paper proposed a stream-based multi-engine multi-channel identification mechanism by analyzing the mobile Internet data packet encapsulation principle, the representation of the character of the word in application layer, the position of the law, the port of the agreement. And the introduction of the adaptive priority matching algorithm reduces the number of packet pattern matching. These features improve the ability of concurrent data processing and efficiency of applications identification. Build scalable dynamic business recognition library that implements the system and applications decoupling. Multi-engine multi-channel identification mechanism integrated existing mature recognition technology, can select the serial, parallel, hybrid, a combination ways to connect the recognition engines, avoid the blind spot of recognition technology cause the recognition failure, make up for the limitations, unity of the present application identification technology. At the same time, multiple guarantee the recognition accuracy, multi-engine multi-channel identification mechanism is also better than the single-engine recognition in concurrent processing, the capacity of data quickly digest has been increased.This paper implements some modules in multi-engine, multi-channel identification subsystem, include table lookup unit, adaptive priority match algorithm and file write module. Use the captured data packets from the mobile network as the experimental data to verify the feasibility and reliability of the modules.