Introduction

In traditional wireless communication systems, localizing a transmitter is relatively straightforward when the structure of the transmitted signals is known. Receivers at different locations can synchronize with signals, decode them, and send key data, like timestamps and signal strength, to a central server. This server then processes the data to pinpoint the transmitter’s location using triangulation or other methods.

However, in scenarios where the signal structure is unknown, such as with rogue transmitters, these techniques become much more challenging. Current approaches, like streaming raw signal data (I/Q data) to a central processor for cross-correlation, are impractical due to the high data volume and computational power required.

The aim of this project is to explore an alternative approach: detecting unknown signals and extracting lightweight features such as transmission length, phase shifts, and bandwidth. These features can then be processed with the goal of enabling accurate localization without the need for high-bandwidth data transfers and processing. By leveraging Software-Defined Radios (SDRs), this project seeks to evaluate the feasibility of this method, providing a solution that compresses signal data for efficient collection and localization.

If successful, this project could significantly advance the field of wireless signal localization and provide a practical method for detecting and localizing unknown transmitters.

Project Goals

• Investigate methods for detecting and extracting signal features from unknown transmitters.
• Evaluate the robustness of the features for different wireless technologies and environments.
• Evaluate the feasibility and accuracy of these methods using SDRs.

Tasks

Analysis of the State-of-the-Art

Perform a comprehensive literature review on current methodologies and technologies in RF signal compression and fingerprinting, with focus on technology agnostic and robust methodologies.

System Design and Implementation

Develop and implement the algorithms and models that would compress raw signals coming from the SDRs, and then perform the target positioning using those compressed representations.

Evaluation and Optimization

Evaluate the system’s accuracy, efficiency, and scalability using synthetic or existing RF datasets, and extend it to real-world traces collected with SDRs. Compare then the positioning accuracy to more traditional methods like cross-correlation.

Requirements

• Interest in wireless communication and signal processing.
• Experience with software-defined radios (SDR) and signal analysis is a plus.
• Strong programming skills in languages like Python and/or C++.
• Good understanding of RF and Machine-Learning ecosystems.

This project provides an excellent opportunity to make a meaningful contribution to the field of wireless localization, working with cutting-edge SDR technology. If you’re passionate about RF systems and want to tackle real-world challenges, we encourage you to apply!