Telecom Anomaly Detection via Call Duration Bins
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Telecommunication carriers face challenges in maintaining guaranteed Quality of Service (QoS) due to inadvertent anomalies in their networks, which can lead to unforeseen decreases in service quality, prompting customers to seek alternative carriers.
Innovation Solution
The method involves analyzing call detail records (CDRs) to detect anomalies by calculating call duration, grouping calls based on duration bins, calculating summary statistics, and using Local Outlier Factor (LOF) analysis to identify potential issues, such as premature termination of conference calls caused by session timers, and generating external tickets to inform carriers of anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If carriers implement cost-reducing features in the network, then operational cost is reduced, but service quality may deteriorate due to inadvertent interference
Solution Approach 1:
The system performs preliminary analysis of call detail records to detect anomalies before they significantly impact service quality. By calculating call durations, grouping calls into duration bins, and computing summary statistics in advance, the system identifies potential issues with cost-reducing features before they cause customer churn, allowing proactive remediation
Solution Approach 2:
The system establishes a feedback loop by continuously monitoring call duration statistics and generating anomaly detections. When anomalies are detected (such as unexpected call termination patterns), the system generates tickets to inform carriers, creating a feedback mechanism that allows carriers to adjust their cost-reducing features to maintain service quality
2Reliability
If the network monitors call details continuously to detect anomalies, then service quality is improved, but system complexity increases
Solution Approach 1:
The system segments the monitoring process into distinct stages: extracting call detail records, calculating call durations, grouping calls into duration bins, computing summary statistics for each bin, and detecting anomalies. This segmentation allows the complex monitoring task to be broken down into manageable, independent processing steps that can be implemented and maintained more easily
Solution Approach 2:
The system introduces summary statistics as an intermediary layer between raw call detail records and anomaly detection. Instead of directly analyzing individual call records, the system aggregates data into summary statistics (such as average call duration per duration bin), which simplifies the anomaly detection process while maintaining effective monitoring capability
Data Source
AI summary
Methods for detecting anomalies in a telecommunications network are provided. The methods include calculating a call duration for each of a plurality of calls associated with a corresponding call detail record (CDR). Each CDR provides details associated with a single call in a plurality of fields. Each of the plurality of calls are categorized based on the calculated call duration for each call such that each of the plurality of calls is assigned to a duration bin, each duration bin including calls having similar calculated call durations. The CDRs are grouped using the duration bins and at least one carrier and/or a telecommunications provider. A plurality of summary statistics are calculated for each of the plurality of CDR groups. The calculated plurality of summary statistics are analyzed to determine whether an anomaly is present. Among other remedial measures, an external ticket may be generated informing affected carriers when an anomaly is present.