Hello.

This post details how I developed a Call Detail Record (CDR) generator for my final year project on telecoms fraud. The generator creates realistic CDRs using a simplified data model, focusing on caller, callee, call type, start time, and duration. Various user models (high, low, business, etc.) are configured with parameters like average call cost, standard deviation, and call frequency for different call types (local, national, etc.), along with likely call times. Random numbers are then generated within these parameters to create a diverse set of CDRs that accurately reflect the modeled behavior.

I've shared my Call Detail Record (CDR) Generator, a Microsoft Access 2003 application I developed for my final year project. This tool creates thousands of simulated phone calls, mimicking various customer profiles (e.g., high/low usage, fraudulent, national rate). It offers extensive customization options for call cost, duration, and user behavior patterns (e.g. business vs. home user calling time). The generated call data is designed for training neural networks. Feel free to download and experiment – it's easy to use.