Network Outage Prediction Using Metrics and Event Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network outage prediction methods are time-consuming and require manual intervention, leading to increased service delivery costs and downtime, and traditional linear regression algorithms oversimplify complex network issues.
Innovation Solution
A method utilizing system metrics and event messages with machine learning techniques to predict network outages by combining time-series data and event data into a unique set of features for accurate outage prediction, allowing proactive network modifications to prevent downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional linear regression algorithms are used for network outage prediction, then the prediction process is simple, but the prediction accuracy is insufficient due to oversimplification of complex network issues
Solution Approach 1:
The patent transforms the prediction approach by changing from simple linear parameters to complex non-linear parameters through embedding techniques. The embedding layer converts categorical network variables into continuous vector representations, capturing complex relationships that linear regression cannot detect, thereby improving prediction accuracy while maintaining computational feasibility.
Solution Approach 2:
The patent creates a composite prediction model that combines multiple components: embedding layers for categorical data, convolutional neural networks for temporal patterns, and dense layers for final prediction. This composite architecture integrates different processing capabilities to handle the multifaceted nature of network outage prediction, overcoming the limitations of single-method approaches.
2Ease of operation
If manual intervention is used for network outage redressal, then the problem can be addressed with human judgment, but the process is time-consuming and increases service delivery costs
Solution Approach 1:
The patent implements a self-service prediction system that automatically monitors network metrics, processes data through the neural network model, and generates outage predictions without human intervention. The system autonomously identifies patterns and predicts outages, eliminating the time-consuming manual analysis phase while maintaining expert-level detection capabilities.
Solution Approach 2:
The patent performs preliminary action by predicting network outages before they occur. The model analyzes historical and real-time data to forecast potential failures, enabling proactive maintenance scheduling and resource allocation. This advance prediction eliminates the need for reactive manual intervention, reducing both response time and operational costs.
3Measurement precision
If network monitoring data is collected and analyzed continuously, then accurate outage prediction can be achieved, but the computational load and processing time increase
Solution Approach 1:
The patent extracts only the most relevant features from the continuous stream of network monitoring data using the embedding layer and convolutional operations. Instead of processing all raw data points, the model identifies and extracts key temporal patterns and anomalies, significantly reducing computational load while maintaining prediction accuracy by focusing on discriminative features.
Data Source
AI summary
An exemplary method, for preventing an outage in a network, includes collecting network metrics and producing statistical features of the network during a predetermined time window based on statistical analysis of the collected metrics. The method further includes selecting, from the produced statistical features, a first data set of relevant statistical features. Further, the method includes collecting information about a past event, extracting event data from the past event, and vectorizing the extracted event data to form a second data set. The method further includes concatenating the first data set and the second data set to form a third data set and classifying the third data set relative to a model of historical network performance to determine a probability of the outage. Further, the method includes modifying, in response to the probability being above a threshold, the network/activity within the network to avoid onset of the outage.