Radio frequency band segmentation, signal detection and labelling using machine learning
Inventors
West, Nathan • Roy, Tamoghna • O'Shea, Timothy James • Hilburn, Ben
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12373715-B2
Publication Date
2025-07-29
Expiration Date
Abstract
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for radio frequency band segmentation, signal detection and labelling using machine learning. In some implementations, a sample of electromagnetic energy processed by one or more radio frequency (RF) communication receivers is received from the one or more receivers. The sample of electromagnetic energy is examined to detect one or more RF signals present in the sample. In response to detecting one or more RF signals present in the sample, the one or more RF signals are extracted from the sample, and time and frequency bounds are estimated for each of the one or more RF signals. For each of the one or more RF signals, at least one of a type of a signal present, or a likelihood of signal being present, in the sample is classified.
Core Innovation
The invention relates to machine-learning-based radio spectrum sensing that receives a sample of electromagnetic energy from one or more RF communication receivers and examines the sample to detect one or more RF signals present in the sample. In response to detecting the RF signals, the method extracts the one or more RF signals from the sample and generates time-frequency bins by estimating time and frequency bounds for each RF signal. The approach supports classifying, for the time-frequency bins, at least one of a signal type being present or a likelihood of being present in at least a plurality of time-frequency bins.
For the time-frequency bins, the invention obtains a signal class probability for each time-frequency bin indicating a signal type being present or likelihood of being present. Based on the signal class probabilities, the method clusters the plurality of time-frequency bins into one or more bounding boxes having a time dimension and a frequency dimension. Each bounding box represents a contiguous portion of a time-frequency region that includes at least one RF signal of a signal type corresponding to at least one time-frequency bin.
The invention further provides a user interface that displays representations of the one or more bounding boxes and an adjacent table with entries providing information on the displayed representations. At least one table entry corresponds to a bounding box and includes fields displaying one or more signal types associated with the respective bounding box, one or more frequency values corresponding to the one or more signal types, a bandwidth corresponding to the one or more signal types, and a received signal strength indicator (RSSI) associated with the respective bounding box.
Claims Coverage
The independent claims are clm-00001 (method) and clm-00018 (system). Both independent claims share the same core inventive workflow and output: detect RF signals in an electromagnetic-energy sample, generate time-frequency bins with estimated time and frequency bounds, classify signal type and/or likelihood per bin using signal class probabilities, cluster bins into time-frequency bounding boxes based on those probabilities, and display bounding boxes with an adjacent table containing signal-related fields.
Processor-based method for RF signal detection, binning, classification, and time-frequency bounding box display
Receives a sample of electromagnetic energy from one or more RF communication receivers, detects one or more RF signals present in the sample, extracts the RF signals from the sample, generates time-frequency bins by estimating time and frequency bounds for each RF signal, classifies each time-frequency bin by obtaining a signal class probability indicating a signal type and/or likelihood of being present, clusters the time-frequency bins into one or more time-frequency bounding boxes based on the signal class probabilities, and displays representations of the one or more bounding boxes with an adjacent table including signal types, frequency values, bandwidth, and RSSI for corresponding bounding boxes.
System for RF signal detection, binning, classification, clustering, and UI table display
Includes one or more processors and computer-readable media storing instructions that receive a sample of electromagnetic energy from an RF communication receiver, detect one or more RF signals present in the sample, generate time-frequency bins by estimating time and frequency bounds for each RF signal, classify time-frequency bins by obtaining signal class probability for each bin indicating signal type and/or likelihood of being present, cluster the plurality of time-frequency bins based on the signal class probabilities into one or more time-frequency bounding boxes, and display representations of the one or more bounding boxes and an adjacent table with fields for signal types, frequency values, bandwidth, and RSSI associated with respective bounding boxes.
Across the independent claims, the inventive coverage centers on producing time-frequency bounding boxes from signal class probabilities derived from classified time-frequency bins, and presenting those bounding boxes with an adjacent table that reports signal type, frequency values, bandwidth, and RSSI.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?