DTMF Call Classification for Spoofing and Device Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing DTMF tone detection systems struggle with deviations due to device type and geographic location, leading to inaccuracies in identifying the device type and authenticity of phone calls, especially in IVR systems.
Innovation Solution
A computer-implemented method that analyzes DTMF information to determine a feature vector, compares it to a model, and classifies calls based on device type and geographic location, accounting for channel and additive noise, using statistical features like mean, median, and variance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If DTMF tone detection is used to identify device type and geographic location, then call classification capability is improved, but accuracy deteriorates due to variations in DTMF tone generation across different devices and geographical locations
Solution Approach 1:
The system performs preliminary actions by collecting DTMF tone samples from multiple devices and locations before actual call classification. These samples are used to pre-compute statistical entities and train classification models, enabling the system to adapt to device-specific variations before they affect classification accuracy
Solution Approach 2:
The system changes parameters by computing multiple statistical entities (mean, median, variance, skewness, kurtosis) from DTMF tone frequencies and durations. These statistical parameters transform raw DTMF variations into standardized features that can be reliably classified, converting device-specific noise into meaningful classification data
2Measurement precision
If statistical entities are computed from DTMF tones to improve device identification, then measurement precision is improved, but device complexity increases due to multiple statistical calculations
Solution Approach 1:
The system segments the complex task of device identification into multiple independent statistical calculations. Each statistical entity (mean, median, variance, etc.) is computed separately from DTMF tone parameters, and each contributes independently to the final classification decision, making the complex process manageable and interpretable
3Reliability
If DTMF variations are analyzed to detect spoofing, then reliability is improved, but difficulty of detecting and measuring increases due to subtle tone modifications
Solution Approach 1:
The system transforms subtle DTMF variations into amplified statistical parameters. By computing statistical entities like variance and kurtosis from DTMF frequencies and durations, the system converts subtle tone modifications into pronounced statistical deviations that are easier to detect and measure for spoofing identification
Data Source
AI summary
Systems, methods, and computer-readable media for call classification and for training a model for call classification, an example method comprising: receiving DTMF information from a plurality of calls; determining, for each of the calls, a feature vector including statistics based on DTMF information such as DTMF residual signal comprising channel noise and additive noise; training a model for classification; comparing a new call feature vector to the model; predicting a device type and geographic location based on the comparison of the new call feature vector to the model; classifying the call as spoofed or genuine; and authenticating a call or altering an IVR call flow.