Reconstruction Engine for Fraud Detection in Distributed Data Centers

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The distribution of data across multiple data centers makes it challenging to obtain a complete representation of a user's activity, hindering fraud detection and system performance analysis, as existing fraud detection systems struggle to collect data in real time without impacting customer experience.

Innovation Solution

A method and system for reconstructing data packets from multiple data centers into a unified format, using a reconstruction engine that processes data packets from various data centers to form a complete user session, which can then be analyzed for fraudulent activity using profiling and rule engines.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If data is distributed across multiple data centers to handle increased traffic and improve processing speed, then system throughput and response time are improved, but the ability to obtain a complete representation of user activity for fraud detection deteriorates

Engineering Contradiction:

Improvedata processing speedVSAvoidcompleteness of user activity data

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The system segments user activity data collection by routing different portions of data packets to different data centers for parallel processing, then uses a reconstruction engine to reassemble the complete user session by matching data packets from multiple data centers using correlation identifiers

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

A reconstruction engine acts as an intermediary component that receives data packets from multiple data centers, correlates them using unique identifiers, and reconstructs complete user sessions, enabling fraud detection systems to access holistic user activity data without impacting customer experience

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If fraud detection systems attempt to collect data from all data centers in real time to detect fraudulent activity, then fraud detection capability is improved, but customer experience deteriorates due to system impact

Engineering Contradiction:

Improvefraud detection accuracyVSAvoidcustomer experience

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The system extracts a representative sample of data packets from the data stream at data centers and routes them to the reconstruction engine for fraud analysis, rather than collecting all data, thereby enabling effective fraud detection while minimizing impact on customer experience

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

Data packets are tagged with correlation identifiers during the normal data flow before reaching the fraud detection system, enabling the reconstruction engine to efficiently assemble complete user sessions without requiring real-time interception or slowing of data transmission

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS8180873B2Detecting fraudulent activity

Publication Date: 2012.05.15 FMR CORP

AI summary

Described are computer-based methods and apparatuses, including computer program products, for subscribing to data feeds on a network. An user utilizes a transmitting device to transmit data packets that are split between a plurality of data centers for processing. The data packets are captured at the data centers. The data packets are recombined. The recombined data can be analyzed for fraud detection, marketing analysis, network intrusion detection, customer service analysis, and/or performance analysis. If fraudulent activity is detected, then the user can be interdicted to prevent further fraudulent activity.