At the Signals Lab, we have developed a variety of innovative platforms to support our cutting-edge research. These platforms are versatile, deployable on heterogeneous hardware, and designed to advance our understanding and management of wireless signals. Leveraging our testbeds, our team and collaborators utilize these tools to push the boundaries of spectrum sensing and signal analysis.
Any software or hardware wireless system can be seamlessly integrated with our platforms on our testbed. To try our your algorithms or capabilities in our facility, please get in touch.
Signal Detection and Summarization
We employ advanced signal detection and summarization techniques to identify and classify various signals in the spectrum. Our tools, such as SparSDR, Searchlight, and Dovetail, are at the forefront of this effort.
Perform blind energy detection and compression of I/Q samples. It compresses I/Q samples in a sparsity-proportional manner, making it highly efficient in terms of resource usage. SparSDR allows for the capture of signals across wide bandwidths by adding lightweight frontend processing and corresponding backend reconstruction to restore signals to their original sample rate.
Searchlight
An upgraded system that builds on SparSDR’s capabilities. It performs blind energy detection and direct I/Q extraction. Written entirely as software on CPU and GPU, Searchlight offers accurate, sensitive, and fast detection of radio frequency energy, identifying the center frequency and reference time of the energy, and drawing bounding boxes around detected signals.
Dovetail
Designed for blind energy detection and activity recognition. It can form patterns for different activities in the spectrum and decompose those activities. For example, Dovetail can separate different signals in the same band, such as WiFi and Bluetooth, and even distinguish between multiple WiFi devices. This tool is invaluable for summarizing signals in the spectrum and analyzing activity.
Signal Visualization
We use advanced visualization tools to explore and interpret wireless signal data. These tools help us understand the behavior of different signals and identify patterns and anomalies in the spectrum.
Visum
RFSynth: Data Generation and Testing Tools
RFSynth is our comprehensive platform for generating and testing RF data. It is designed to address the challenges of limited data generation and testing platforms for spectrum sensing systems. The framework combines simulation and real-world data generation methods to enable large and diverse datasets for training and testing RF ML models and spectrum sensing systems.
RFSynth includes modules for metadata generation, allowing for easy experimentation. The effectiveness of the framework is demonstrated through experiments, including signal detection and modulation classification. This platform contributes to the development of a scalable framework for generating RF IQ data with ease, significantly reducing the development and deployment time of complex wireless systems.
Software Digital Signal Processing Libraries
Our software digital signal processing libraries provide essential tools for processing and analyzing wireless signals. These libraries are optimized for performance and can be deployed on various hardware platforms. Most of our code is written in C++ and Rust.
We have software implementations of Polyphase filter bank channelizer and synthesis, STFT channelizer and synthesis on GPU, and efficient implementations of FFT resampling. These tools allow for the efficient processing of wideband signals and the extraction of meaningful information from the spectrum.
We also have GPU-accelerated implementations of Polyphase filter bank channelizer and synthesis (in CUDA C++), and the Strip Spectral Correlation Analyzer (SSCA) for cyclostationary features. These implementations leverage the computational power of GPUs to provide fast and accurate signal processing capabilities.
Spectrum Sensing Hardware
Our spectrum sensing hardware includes state-of-the-art Software-Defined Radios (SDRs) such as the USRP N320 and Signal Hound SM200C. These devices allow us to capture and analyze wireless signals with high fidelity.
We also employ sweeping spectrum sensors like SweepSense and Crescendo, which enable continuous monitoring and detection of signals across a wide frequency range. These sensors enhance our ability to perform real-time spectrum analysis and detect transient signals that other systems might miss.