Automated DPI Rule Adaptation for Network Service Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The reliability of service identification in cellular networks using Deep Packet Inspection (DPI) is hindered by the laborious process of manually developing and updating DPI rules, which is inefficient for the large and increasing number of mobile services, limiting scalability and accuracy.
Innovation Solution
A method for maintaining DPI rules that allows for automated association and adaptation based on reports from a subset of mobile terminals, enabling effective identification of network services without manual training data models, and scaling with the number of services and terminals, reducing network traffic and improving identification rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual DPI rule development is used, then rule accuracy can be achieved, but the process is laborious and cannot scale to thousands or millions of client applications
Solution Approach 1:
The system enables self-service by allowing the network to automatically learn and maintain DPI rules through machine learning algorithms. The network node autonomously collects data packets, identifies service patterns, and updates rules without manual intervention, resolving the contradiction between accuracy and scalability.
Solution Approach 2:
The patent replaces the mechanical manual rule development process with automated machine learning systems. Instead of manual analysis and rule creation, the system uses algorithms to automatically learn service patterns from data packets and generate DPI rules, dramatically improving productivity while maintaining reliability.
2Reliability
If DPI rules are manually maintained, then identification accuracy can be preserved, but the process cannot keep up with the increasing number of new services and application behavior changes
Solution Approach 1:
The system implements dynamics by enabling continuous adaptation of DPI rules through ongoing machine learning processes. The network node continuously learns from new data packets and updates rules to accommodate emerging services and changing application behaviors, making the system both reliable and highly adaptable.
Solution Approach 2:
The patent incorporates feedback mechanisms where the system continuously monitors service identification results and uses this information to refine and update DPI rules. This closed-loop approach ensures high reliability while automatically adapting to new services without manual intervention.
3Manufacturing precision
If extensive data collection from many mobile terminals is performed, then comprehensive rule training is possible, but network traffic overhead increases significantly
Solution Approach 1:
The system applies partial action by collecting data packets from a selective subset of mobile terminals rather than all terminals. The machine learning algorithm processes packets from this representative sample to learn service patterns, achieving comprehensive rule training with reduced network traffic overhead.
Solution Approach 2:
The patent extracts only the necessary data packets needed for rule learning from the network traffic. Instead of collecting all traffic from all terminals, the system selectively extracts relevant packets from a subset of terminals, reducing overhead while maintaining training comprehensiveness.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A technique for maintaining rules for identifying network services is presented. The network services exchange data packets (136, 138) with mobile terminals (108, 110) wirelessly connected to a network (100). The network receives reports from a subset (132) of the mobile terminals. The reports indicate network services used by the reporting mobile terminals. As to a method aspect of the technique, the network associates one or more first data packets (136) with one of the network services based on the reports. The first data packets originate from, or are addressed to, the reporting mobile terminals. The network derives or adapts, based on data included in the first data packets, one or more rules for identifying the network service based on data included in second data packets (138). The second data packets are different from the first data packets.