ML Alarm Suppression for Network Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Network operators face challenges in managing complex multi-layer networks with numerous alarms, requiring expertise to differentiate between important and non-actionable alerts, as existing systems rely on manual intervention and static rules that are cumbersome to maintain and fail to adapt to evolving networks.
Innovation Solution
The implementation of machine learning models using supervised and reinforcement learning techniques to automatically classify alarms as important or non-important, trained with historical data and NOC interactions, allowing for real-time suppression of non-actionable alarms while maintaining a high recall of important ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual intervention and static rules are used to manage alarms, then network operators can differentiate between important and non-actionable alerts, but the system becomes cumbersome to maintain and fails to adapt to evolving networks
Solution Approach 1:
The patent implements dynamic alarm suppression by replacing static rules with machine learning models that continuously learn from new data. The system adapts to evolving network patterns by training on historical alarm data and updating suppression decisions in real-time, allowing the alarm management system to evolve with the network without requiring manual rule updates.
Solution Approach 2:
The machine learning model performs self-service by automatically learning from historical alarm data and NOC interactions without requiring manual intervention. The system trains itself on past alarm patterns and autonomously improves its classification accuracy, eliminating the need for network operators to continuously maintain and update static rules.
2Loss of information
If all alarms are displayed to network operators, then complete information is available for decision-making, but the volume of alarms becomes unmanageable and reduces operational efficiency
Solution Approach 1:
The system extracts and suppresses non-actionable alarms from the overall alarm stream based on machine learning classification. By identifying and removing redundant alarm types (such as recurring non-critical alarms) from the displayed alarm list, the system maintains complete information for actionable alarms while filtering out noise that reduces operational efficiency.
Solution Approach 2:
The system applies partial suppression by selectively hiding only the non-actionable portion of alarms while maintaining full visibility of actionable alarms. This partial action approach ensures that operators receive sufficient information to make decisions without being overwhelmed by the excessive volume of all alarms, achieving the right balance between information completeness and operational efficiency.
3Productivity
If machine learning models suppress non-important alarms, then alarm volume is reduced by 50%, but there is risk of missing important actionable alarms
Solution Approach 1:
The system implements feedback loops where NOC operator interactions with suppressed alarms are captured and used to retrain the machine learning model. When operators take actions on suppressed alarms or provide feedback, this information feeds back into the training data, allowing the model to learn from real-world outcomes and continuously improve its accuracy in distinguishing important from non-important alarms, thereby maintaining high reliability.
Solution Approach 2:
The machine learning model performs preliminary classification and suppression of non-important alarms before they reach operators. By pre-filtering alarms based on learned patterns from historical data, the system reduces alarm volume to 50% while maintaining high detection accuracy for important alarms, as the model has already performed the initial sorting and prioritization work.
4Ease of operation
If device-level alarm suppression is implemented, then specific alarm types can be suppressed, but critical alarms such as link down and power failure cannot be suppressed
Solution Approach 1:
The machine learning model provides a universal alarm suppression mechanism that works across all alarm types and device levels without requiring device-specific configuration. The single model handles diverse alarm categories (link down, power failure, application-level alarms) uniformly by learning from aggregated historical data, eliminating the need for separate suppression configurations for different device types and alarm categories.
Data Source
AI summary
Systems and methods include receiving alarms from a network; utilizing a machine learning model to classify the alarms as one of important and non-important; and displaying the important alarms and suppressing display of the non-important alarms. The systems and methods can further include training the machine learning model with historical alarm data that includes features related to an associated device and comments related to how a Network Operations Center (NOC) handles an associated alarm or group of alarms. The training can be via supervised machine learning with the features used as labels or via reinforcement learning with the features used as a reward.