Trained Data Model for Automated Network Alarm Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual process of interpreting and addressing network alarms in computer networks is cumbersome due to the need for technical expertise and the complexity of managing multiple Management Information Bases (MIBs) from various vendors, leading to inefficiencies and potential human errors in fault identification and resolution.
Innovation Solution
A system utilizing a trained data model that learns from multiple network alarms from diverse devices to analyze and map network alarms with standard attributes, enabling automated recognition and addressing of network faults.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual interpretation of network alarms is used, then technical expertise can be applied to analyze faults, but the process becomes cumbersome and time-consuming
Solution Approach 1:
The patent replaces the manual mechanical process of interpreting network alarms with an automated machine learning system. The trained data model automatically analyzes alarm information, extracts fault patterns, and generates diagnostic results without human intervention, thereby reducing time loss while maintaining identification accuracy through systematic algorithmic processing.
Solution Approach 2:
The system enables self-service by allowing the network management system to automatically diagnose and identify faults without requiring manual technical expertise. The trained data model independently processes alarm information, performs pattern recognition, and generates fault identification results, making the system self-sufficient in fault detection tasks.
2Adaptability or versatility
If manual management of multiple MIBs from various vendors is performed, then comprehensive device monitoring is achieved, but complexity increases and human errors occur
Solution Approach 1:
The patent implements a universal trained data model that can handle multiple MIBs from various vendors through a single system. The model is trained on diverse alarm data from different device types and vendors, enabling it to universally process and interpret alarms across multi-vendor environments without requiring separate management approaches for each vendor's MIBs.
Solution Approach 2:
The trained data model acts as an intermediary layer between the diverse vendor-specific MIBs and the network management system. It standardizes the processing of different MIB formats and alarm structures by translating them into a unified analysis framework, thereby reducing the complexity of directly managing multiple vendor-specific MIBs.
3Productivity
If automated systems are used to reduce manual effort, then efficiency improves, but accuracy in interpreting complex network alarms may deteriorate
Solution Approach 1:
The system performs preliminary action by training the data model extensively on historical alarm data before deployment. This pre-training phase enables the automated system to learn complex alarm patterns, correlations, and fault signatures in advance, ensuring that when the system operates in production, it can accurately interpret alarms without requiring manual oversight, thus maintaining both high productivity and precision.
Data Source
AI summary
The present invention relates a system and a method of recognizing and addressing network alarms in a computer network. A network adapter is configured to receive network alarms related to operating condition of network devices present in the computer network, wherein the network devices are managed by different vendors. Information present in the network alarms is analyzed to determine elements indicating the operating condition of the network devices. The elements may comprise at least one of keywords, object identifiers, and values of the object identifiers. A trained data model is finally used for mapping the network alarms with standard attributes. Based on such mappings, the network alarms are then addressed.