Fault Escalation Engine Using SVM Alarm Categorization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fault management systems in large networks struggle to accurately escalate alarms due to hard-coded rules, often missing critical alarms and escalating unnecessary ones, especially in complex environments with numerous network devices.
Innovation Solution
Implementing a fault escalation engine that utilizes machine learning techniques, such as Support Vector Machines and statistical analysis, to intelligently categorize alarms based on persistence, repetition, and recovery, allowing for more accurate escalation decisions with a statistically meaningful confidence level.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If hard-coded rules are used to automate fault management, then automation extent is improved, but measurement precision of fault escalation decisions deteriorates
Solution Approach 1:
The patent replaces hard-coded rule-based systems with machine learning models that learn optimal escalation thresholds from historical alarm data. The system uses supervised learning algorithms to train classifiers that predict whether alarms require escalation, substituting rigid mechanical rules with adaptive intelligent systems that continuously improve decision accuracy based on accumulated experience.
Solution Approach 2:
The system dynamically adjusts escalation parameters such as time windows, alarm counts, and severity thresholds based on learned patterns from historical data. Instead of fixed parameters, the machine learning models optimize these parameters adaptively, changing them according to network conditions, alarm types, and temporal patterns to maintain high precision in escalation decisions.
2Device complexity
If hard-coded rules with fixed thresholds are used, then device complexity is reduced, but reliability of fault detection deteriorates
Solution Approach 1:
The system performs preliminary training using historical alarm data before deployment, pre-learning optimal detection patterns and thresholds. This preliminary action allows the system to encode sophisticated detection logic during the training phase, so that during operation, the actual detection process remains relatively simple while maintaining high reliability through the pre-acquired knowledge.
Solution Approach 2:
The patent creates a virtual model of normal and abnormal network conditions by training on historical data. This copied knowledge from historical patterns is then applied to current alarm evaluation, allowing the system to leverage past experiences without requiring complex real-time analysis of every individual alarm, thus maintaining simplicity while improving reliability.
Data Source
AI summary
A processing system includes a memory storing a program of instructions, and a processor coupled to the memory and configured to execute the program of instructions. The processor monitors a network, which includes a plurality of devices, for alarms generated by the plurality of devices. The alarms are associated with alarm conditions, and the processor assigns the alarm conditions associated with received alarms to escalation-type categories based on Support Vector Machine learning analysis of alarm characteristics of the alarm conditions. The processor also determines whether a first alarm condition associated with a first received alarm satisfies a first escalation criterion associated with a first escalation-type category to which the first alarm condition has been assigned, and triggers an escalation action in response to determining that the first alarm condition satisfies the first escalation criterion.