SIM Fraud Detection via IMSI Stability History
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing SIM card fraud detection systems rely on simple IMSI checks, leading to high false positives and increased call centre costs due to their inability to quantify the probability of fraud, causing legitimate SIM changes to be treated as potential fraud and resulting in poor customer experience.
Innovation Solution
A SIM fraud detection system that maintains historical records of IMSI changes and uses statistical methods to estimate the probability of fraud by analyzing IMSI stability, incorporating MCC/MNC information and caching pre-calculated data for real-time on-demand checks, differentiating between fraudulent and non-fraudulent changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a simple IMSI check is used to detect SIM fraud, then the detection method is easy to implement, but it produces high false positives and cannot quantify the probability of fraud
Solution Approach 1:
The system performs preliminary background checks on IMSI changes by maintaining historical records and pre-calculating statistical data. This preliminary action enables the system to provide quantitative fraud probability assessments in real-time without requiring complex calculations at the moment of detection, thus maintaining ease of implementation while improving accuracy.
Solution Approach 2:
The system uses feedback from historical IMSI change data to continuously refine its statistical models. By analyzing past patterns and updating probability assessments based on observed behavior, the system improves its fraud detection accuracy while keeping the implementation straightforward through automated learning processes.
2Reliability
If every IMSI change is treated as potential fraud requiring manual intervention, then fraud detection reliability is improved, but call centre costs increase and customer experience deteriorates
Solution Approach 1:
The system changes the parameter from binary fraud detection (fraud or not fraud) to quantitative probability assessment (0-100% fraud likelihood). This parameter change enables differentiated responses: high-probability changes trigger manual review while low-probability changes can be automatically handled, thus maintaining reliability while improving call centre efficiency.
Solution Approach 2:
The system dynamically adjusts the level of manual intervention based on real-time probability assessments. Instead of a static all-or-nothing approach, the system adapts its response strategy based on the calculated fraud probability, allowing automated handling of legitimate changes while maintaining manual review for suspicious cases.
3Reliability
If manual intervention is required for all SIM changes, then fraud detection thoroughness is improved, but customer experience is downgraded and costs increase
Solution Approach 1:
The system enables self-service handling of legitimate SIM changes through automated probability assessment. When the system determines low fraud probability, it automatically allows the SIM change without requiring customer intervention, thus maintaining thoroughness while significantly improving customer experience and reducing costs.
4Measurement precision
If historical IMSI data is maintained and analyzed statistically, then fraud probability quantification is improved, but system complexity increases
Solution Approach 1:
The system performs preliminary data collection and statistical pre-calculations during background checks. By preparing historical data and statistical models in advance, the system can provide accurate fraud probability quantification during real-time detection without requiring complex on-demand calculations, thus managing complexity effectively.
Solution Approach 2:
The system creates a simplified representation of complex fraud patterns through statistical models and historical data summaries. This copying approach allows the system to handle complex fraud detection problems through manageable statistical frameworks rather than complex real-time analysis algorithms.
Data Source
Figure 1
AI summary
A SIM fraud detection system (1) stores a current list (2) of monitored MSISDNs and maintains a set of historical check records (3) by performing background checks by querying a mobile network (20) probe to determine the IMSI and MCC/MNC associated with each monitored MSISDN at a point in time and including in historical check records stability flags that describe the relationship of the current record with others for the same MSISDN. The stability data provides an indication of legitimacy of an IMSI change. In real time, the system responds to on- demand checks by estimating the probability that a monitored MSISDN has been fraudulently associated with a mobile device by querying for a current IMSI, MCC and MNC for the MSISDN, using the information in the set of historical check records, including the frequency of similar checks that were or were not followed by a period of IMSI stability, to generate a probability of fraud, and adding an historical check record recording the on-demand check.