Systems, methods, and devices for automatic signal detection based on power distribution by frequency over time

Abstract

Systems, methods, and devices for automatic signal detection in an RF environment are disclosed. A sensor device in a nodal network comprises at least one RF receiver, a generator engine, and an analyzer engine. The at least one RF receiver measures power levels in the RF environment and generates FFT data based on power level data. The generator engine calculates a power distribution by frequency of the RF environment in real time or near real time, including a first derivative and a second derivative of the FFT data. The analyzer engine creates a baseline based on statistical calculations of the power levels measured in the RF environment for a predetermined period of time, and identifies at least one signal based on the first derivative and the second derivative of the FFT data in at least one conflict situation from comparing live power distribution to the baseline of the RF environment.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

This present invention relates to systems and methods for automatic signal detection. More particularly, the systems and methods of the present invention are directed to automatic signal detection based on power distribution by frequency over time.

2. Description of the Prior Art

In recent years, demand for real-time information has increased exponentially. Consumers have embraced social media applications and there are now more mobile subscriptions than people on the planet. Studies show that a typical mobile device experiences an average of 10 network interactions per minute (e.g., Facebook push, Twitter download). For example, Facebook on its own is driving 1 billion updates per minute. Rabid consumer demand, combined with the growing needs of government and industry (e.g., 2-way, trunked, IoT), translates into more wireless activities over wider frequency ranges. The activities are often intermittent with short durations of only a few hundred milliseconds. Social media applications and other cellular activities (e.g., background refresh) are even shorter in duration. Until now, the magnitude of activity has been impossible to keep track of and even harder to gain intelligence from.

SUMMARY OF THE INVENTION

The present invention provides systems and methods for automatic signal detection based on power distribution by frequency over time, especially based on change of power and rate of change of power over time.

These and other aspects of the present invention will become apparent to those skilled in the art after a reading of the following description of the preferred embodiment when considered with the drawings, as they support the claimed invention.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a configuration of a PDFT processor according to one embodiment of the present invention.

FIG. 2 is a flow chart for data processing in a PDFT processor according to one embodiment of the present invention.

FIG. 3 illustrates data analytics in an analyzer engine according to one embodiment of the present invention.

FIG. 4 illustrates a mask according to one embodiment of the present invention.

FIG. 5 illustrates a workflow of automatic signal detection according to one embodiment of the present invention.

FIG. 6 is a screenshot illustrating alarm visualization via a graphical user interface according to one embodiment of the present invention.

FIG. 7 illustrates a comparison of live FFT stream data and a mask considering a db offset according to one embodiment of the present invention.

FIG. 8 is a snippet of the code of the detection algorithm defining a flag according to one embodiment of the embodiment.

FIG. 9 is a snippet of the code of the detection algorithm identifying peak values according to one embodiment of the present invention.

FIG. 10 illustrates a complex spectrum situation according to one embodiment of the present invention.

FIG. 11 is an analysis of the live stream data above the mask in the first alarm duration in FIG. 10.

FIG. 12 is a snippet of the code of the detection algorithm checking the alarm duration according to one embodiment of the present invention.

FIG. 13 is a snippet of the code of the detection algorithm triggering an alarm according to one embodiment of the present invention.

FIG. 14 is a screenshot illustrating a job manager screen according to one embodiment of the present invention.

FIG. 15 illustrates trigger and alarm management according to one embodiment of the present invention.

FIG. 16 is a screenshot illustrating a spectrum with RF signals and related analysis.

FIG. 17 is a screenshot illustrating identified signals based on the analysis in FIG. 16.

FIG. 18 is a diagram of a modular architecture according to one embodiment of the present invention.

FIG. 19 illustrates a communications environment according to one embodiment of the present invention.

FIG. 20 illustrates an UAS interface according to one embodiment of the present invention.

FIG. 21 lists signal strength measurements according to one embodiment of the present invention.

FIG. 22 illustrates a focused jammer in a mobile application according to one embodiment of the present invention.

FIG. 23 illustrates a swept RF interference by a jammer according to one embodiment of the present invention.

FIG. 24 illustrates data collection, distillation and reporting according to one embodiment of the present invention.

FIG. 25 is a comparison of multiple methodologies for detecting and classifying UAS.

FIG. 26 lists capabilities of an RF-based counter-UAS system according to one embodiment of the present invention.

FIG. 27 illustrates an RF-based counter-UAS system deployed as a long-distance detection model according to one embodiment of the present invention.

FIG. 28 illustrates features of drones in the OcuSync family.

FIG. 29 illustrates features of drones in the Lightbridge family.

FIG. 30 illustrates a spectrum monitoring system detecting an anomalous signal in close proximity of critical infrastructure.

FIG. 31 illustrates a system configuration and interface according to one embodiment of the present invention.

FIG. 32 is a screenshot illustrating no alarm going off for an anomalous signal from LMR traffic not in proximity of site according to one embodiment of the present invention.

FIG. 33 illustrates a GUI of a remote alarm manager according to one embodiment of the present invention.

FIG. 34 labels different parts of a front panel of a spectrum monitoring device according to one embodiment of the present invention.

FIG. 35 lists all the labels in FIG. 33 representing different part of the front panel of the spectrum monitoring device according to one embodiment of the present invention.

FIG. 36 illustrates a spectrum monitoring device scanning a spectrum from 40 MHz to 6 GHz according to one embodiment of the present invention.

FIG. 37 lists the capabilities of a spectrum monitoring system according to 5 main on-network mobile phone states plus 1 no-network mobile phone state.

FIG. 38 illustrates a mobile event analysis per one minute intervals according to one embodiment of the present invention.

FIG. 39 is a site cellular survey result according to one embodiment of the present invention.

DETAILED DESCRIPTION

The present invention provides systems and methods for unmanned vehicle recognition. The present invention relates to automatic signal detection, temporal feature extraction, geolocation, and edge processing disclosed in U.S. patent application Ser. No. 15/412,982 filed Jan. 23, 2017, U.S. patent application Ser. No. 15/681,521 filed Aug. 21, 2017, U.S. patent application Ser. No. 15/681,540 filed Aug. 21, 2017, U.S. patent application Ser. No. 15/681,558 filed Aug. 21, 2017, each of which is incorporated herein by reference in their entirety.

In one embodiment of the present invention, automatic signal detection in an RF environment is based on power distribution by frequency over time (PDFT), including the first derivative and the second derivative values. A PDFT processor is provided for automatic signal detection.

In one embodiment, the PDFT processor increments power values in a 2-Dimensional (2D) array from a frequency spectrum over a set length of time. The length of time is user-settable. For example, the length of time is operable to be set at 5 minutes, 1 hour, or 1 day. The length of time is operable to be set as low as 1 second. Typically, the smallest time interval for setting the environment is 5 seconds. A histogram with frequency as the horizontal axis and power as the vertical axis is operable to be used to describe power values across a spectrum during a certain period of time, which is called the Power Bin Occurrence (PBO).

In one embodiment, power levels are collected for a specified length of time, and statistical calculations are performed on the PBO to obtain the power distribution by frequency for a certain time segment (PDFT). The statistical calculations create baseline signals and identify what is normal in an RF environment, and what are changes to the RF environment. PBO data is constantly updated and compared to baseline to detect anything unique in the RF environment.

The PDFT collects power values and describes the RF environment with collected power values by frequency collected over the time range of the collection. For example, the PDFT processor learns what should be present in the RF environment in a certain area during the time segment from 3 pm to 5 pm. If there is a deviation from historical information, the PDFT processor is configured to send an alarm to operators.

In one embodiment, PBO is used to populate a 3-Dimensional (3D) array and create the Second Order Power Bin Occurrence (SOPBO). The time segment of the PBO is a factor of the length of the SOPBO time segment. The first two dimensions are the same as in PBO, but the third dimension in SOPBO describes how often the corresponding frequency bin and power bin is populated over the SOPBO time segment. The result is operable to be described as a collection of several 2D histograms across a percent of occurrence bins such that each histogram represents a different frequency bin and power bin combination. This provides a percentage of utilization of the frequency for non-constant signals such as RADAR, asynchronous data on demand links or push-to-talk voice.

In one embodiment, the PBO, PDFT, and SOPBO data sets are used for signal detection. For example, statistical calculations of PBOs during a certain time segment are used along with a set of detection parameters to identify possible signals. A frequency-dependent noise floor is calculated by taking the spectral mean from the PDFT data and applying a type of median filter over subsets of frequency. For example, but not for limitation, detection parameters include known signals, basic characteristics, databases of telecom signals, and etc. For example, but not for limitation, median filter types include Standard Median Filter (MF), Weighted Median Filter (WMF), Adaptive Median Filter (AMF) and Decision Based Median Filter (DBMF). The noise floor is then assessed for large changes in power, which indicates the noise floor values are following the curvature of possible signals. At these frequencies, the noise floor is adjusted to adjacent values. Power values below the noise floor are ignored in the rest of the signal detection process. To detect signals, the first derivative is calculated from a smoothed PDFT frequency spectrum. Derivative values exceeding a threshold set based on the detection parameters are matched to nearby values along the frequency spectrum that are equal and opposite within a small uncertainty level. Once frequency edges are found, power values are used to further classify signals. The whole process including the noise floor calculation is repeated for different time segments. The detection parameters are adjusted over time based on signals found or not found, allowing the signal detection process to develop as the PDFT processor runs.

The first derivative of the FFT data is used to detect signals, measure power, frequency and bandwidths of detected signals, determine noise floor and variations, and classify detected signals (e.g., wideband signals, narrowband signals). The second derivative of the FFT data is used to calculate velocity (i.e., change of power) and acceleration (i.e., rate of change of power), and identify movements based on changes and/or doppler effect. For example, the second derivative of the FFT data in an RF environment is operable to be used to determine if a signal emitting device is near road or moving with a car. A SOPBO is the second derivative (i.e., a rate of change of power). It tells us if the signal is varying in time. For example, a simplex network has base station signals transmitting at certain time segments and mobile signals in a different time segment. The SOPBO can catch the mobile signals while the first order PBO cannot. For signals that vary in time such as Time Division Duplex (TDD) LTE or a Radar, SOPBO is important.

FIG. 1 illustrates a configuration of a PDFT processor according to one embodiment of the present invention. In one embodiment, a PDFT processor for automatic signal detection comprises a management plane, at least one RF receiver, a generator engine, and an analyzer engine. The management plane is operable to configure, monitor and manage job functions of the PDFT processor. The at least one RF receiver is operable to receive RF data, generate I/Q data based on the received RF data, and perform FFT analysis. The generator engine is configured to perform a PBO process, and generate PDFT data and SOPBO data based on PBO data. The analyzer engine is configured to calculate noise floor, smooth max hold, generate a PDFT baseline, and identify signals. The smooth max hold function is a curve fitting process with a partial differential equation to provide a running average across adjacent points to reject impulse noise that is operable to be present in the FFT data. The analyzer engine is further configured to calculate a SOPBO baseline based on the SOPBO data.

FIG. 2 is a flow chart for data processing in a PDFT processor according to one embodiment of the present invention. A job manifest is created for initial configuration of a PDFT generator engine or updating the configuration of the PDFT generator engine. The job manifest also starts an RF receiver to receive radio data from an RF environment. The received radio data is transmitted to an FFT engine for FFT analysis. The PDFT generator engine pulls FFT data stream from the FFT engine to build up a based PBO and run a PBO process continuously. An SOPBO process and a PDFT process are performed based on PBO data. SOPBO data from the SOPBO process and PDFT data from the PDFT process is published and saved to storage. The data from the PDFT generator engine is transmitted to an PDFT analyzer engine for analytics including signal detection and classification, event detection and environment monitoring, mask creation, and other analyzer services.

FIG. 3 illustrates data analytics in an analyzer engine according to one embodiment of the present invention. Classical RF techniques and new RF techniques are combined to perform data analytics including environment monitoring and signal classification. Classical RF techniques are based on known signals and initial parameters including demodulation parameters, prior knowledge parameters, and user provided parameters. New RF techniques use machine learning to learn signal detection parameters and signal properties to update detection parameters for signal classification. New signals are found and used to update learned signal detection parameters and taught signal properties based on supervised and unsupervised machine learning.

In one embodiment, the automatic signal detection process includes mask creation and environment analysis using masks. Mask creation is a process of elaborating a representation of an RF environment by analyzing a spectrum of signals over a certain period of time. A desired frequency range is entered by a user to create a mask, and FFT streaming data is also used in the mask creation process. A first derivative is calculated and used for identifying maximum power values. A moving average value is created as FFT data is received during a time period selected by the user for mask creation. For example, the time period is 10 seconds. The result is an FFT array with an average of maximum power values, which is called a mask. FIG. 4 illustrates a mask according to one embodiment of the present invention.

In one embodiment, the mask is used for environment analysis. In one embodiment, the mask is used for identifying potential unwanted signals in an RF environment.

Each mask has an analysis time. During its analysis time, a mask is scanned and live FFT streaming data is compared against the mask before next mask arrives. If a value is detected over the mask range, a trigger analysis is performed. Each mask has a set of trigger conditions, and an alarm is triggered into the system if the trigger conditions are met. In one embodiment, there are three main trigger conditions including alarm duration, db offset, and count. The alarm duration is a time window an alarm needs to appear to be considered as one. For example, the time window is 2 seconds. If a signal is seen for 2 seconds, it passes to the next condition. The db offset is the db value a signal needs to be above the mask to be considered as a potential alarm. The count is the number of times the first two conditions need to happen before an alarm is triggered into the system.

FIG. 5 illustrates a workflow of automatic signal detection according to one embodiment of the present invention. A mask definition is specified by a user for an automatic signal detection process including creating masks, saving masks, and performing environment analysis based on the masks created and FFT data stream from a radio server. If trigger conditions are met, alarms are triggered and stored to a local database for visualization.

FIG. 6 is a screenshot illustrating alarm visualization via a graphical user interface (GUI) according to one embodiment of the present invention. In the GUI, current alarms, acknowledged alarms, and dismissed alarms in a certain RF environment are listed with information including types, counts, durations, carrier frequencies, technologies, and band allocations.

In one embodiment, a detection algorithm is used for alarm triggering. The detection algorithm detects power values over the mask considering the db offset condition, but does not trigger an alarm yet. FIG. 7 illustrates a comparison of live FFT stream data and a mask considering a db offset according to one embodiment of the present invention. The db offset is 5 db, so the detection algorithm only identifies power values that are at least 5 db higher than the mask.

The detection algorithm then identifies peaks for power values above the mask after considering the db offset. In embodiment of the present invention, a flag is used for identifying peak values. A flag is a Boolean value used for indicating a binary choice. FIG. 8 is a snippet of the code of the detection algorithm defining a flag according to one embodiment of the embodiment. If the flag is TRUE, the detection algorithm keeps looking for peak values. A forEach function analyzes each value to find the next peak. Once reaching a peak value, it goes down to the value nearest to the mask, and the flag is set to FALSE. FIG. 9 is a snippet of the code of the detection algorithm identifying peak values according to one embodiment of the present invention.

In one embodiment, live FFT stream data has multiple peaks before falling under the mask. FIG. 10 illustrates a complex spectrum situation according to one embodiment of the present invention. Live FFT stream data in two alarm durations have multiple peaks before falling under the mask. FIG. 11 is an analysis of the live FFT stream data above the mask in the first alarm duration in FIG. 10 according to one embodiment of the present invention. A first peak is identified, and the power value starts to decrease. A first value nearest to the mask after the first peak is identified, the flag is still TRUE after comparing the first value nearest to the mask and mask, so the detection algorithm keeps looking for peaks. Then, a second peak is identified, and the power value starts to decrease. A second value nearest to the mask after the second peak is identified. The second value is greater than the first value, the flag is still TRUE, so the detection algorithm keeps looking for peak values. Then a third peak value is identified and a third value nearest to the mask is also identified. The third value is on the mask considering the offset value, and the flag is set to FALSE. By comparison, the third peak value is considered as the real peak value for the power values above the mask in the first alarm duration of FIG. 10. Once all the peaks are found, the detection algorithm checks the alarm duration, which is a time window where a signal needs to be seen in order to be considered for alarm triggering. The first time that the detection algorithm sees the peak, it saves the time in memory. If the signal is still present during the time window, or appears and disappears during that time, the detection algorithm is to consider triggering an alarm. If the condition is not met, a real-time alarm is not sent to a user, however the detected sequence is recorded for future analysis. FIG. 12 is a snippet of the code of the detection algorithm checking the alarm duration according to one embodiment of the present invention.

If both the db offset condition and the alarm duration condition are met, the detection algorithm analyzes the count condition. If the amount of times specified in the count condition is met, the detection algorithm triggers the alarm. In one embodiment, all alarms are returned as a JSON array, and a forEach function creates the structure and triggers the alarm. FIG. 13 is a snippet of the code of the detection algorithm triggering an alarm according to one embodiment of the present invention.

The present invention provides spectrum monitoring and management, spectrum utilization improvements, critical asset protection/physical security, interference detection and identification, real time situational awareness, drone threat management, and signal intelligence (SigINT). Advantageously, the automatic signal detection in the present invention provides automated and real-time processing, environmental learning, autonomous alarming and operations (e.g., direction finding, demodulation), wideband detection, etc. The automatic signal detection in the present invention is of high speed and high resolution with low backhaul requirements, and can work in both portal and fixed modes with cell and land mobile radio (LMR) demodulation capability. The automatic signal detection system in the present invention is operable to integrate with third party architecture, and is operable to be configured with distributed architecture and remote management. In one embodiment, the automatic signal detection of the present invention is integrable with any radio server including any radio and software defined radio, for example, Ettus SDR radio products.

Specifically, spectrum solutions provided by the automatic signal detection technology in the present invention have the following advantages: task automation, edge processing, high-level modular architecture, and wideband analysis.

Task automation simplifies the work effort required to perform the following tasks, including receiver configuration, process flow and orchestration, trigger and alarm management, autonomous identification of conflicts and anomalous signal detection, automated analytics and reporting, system health management (e.g., system issues/recovery, software update, etc.). FIG. 14 is a screenshot illustrating a job manager screen according to one embodiment of the present invention. FIG. 15 illustrates trigger and alarm management according to one embodiment of the present invention.

Task automation enables an operator to send a job to one or multiple systems distributed across a geography. Each job contains a pre-built, editable manifest, which can configure receivers and outline alarm conditions with appropriate actions to execute. As an example, for a baseline analysis task, the system automatically scans multiple blocks of spectrum in UHF, VHF, Telco bands and ISM bands such as 2.4 GHz and 5.8 GHz, stores multiple derivatives regarding signal and noise floor activity, produces an automated report showing activity and occupancy over a specified time, analyzes signal activity to correctly channelize activity by center frequency and bandwidth, and combines customer supplied or nationally available databases with data collected to add context (e.g., license, utilization, etc.). The baseline analysis task provides an operator with a view into a spectral environment regarding utilization and occupancy. This is operable to be of assistance when multiple entities (local, state and federal agencies) have coverage during a critical event and need to coordinate frequencies. Multiple radios along with multiple systems across a geography are operable to be commanded to begin gathering data in the appropriate frequency bands. Resolution bandwidth and attenuation levels are adjustable, coordination is made simple, and actionable information is returned without significant manual effort.

The systems provided in the present invention is operable to process RF data and perform data manipulation directly at the sensor level. All data is operable to be pushed to a server, but by processing the data first at the sensor, much like in IoT applications, more is operable to be done with less. Overall, edge processing makes information more actionable and reduces cost. The systems of the present invention also leverage machine learning to drive automation at the edge to a higher level, which makes solutions provided by the present invention more intuitive, with greater capability than other remote spectrum monitoring solutions. Edge processing also reduces the bandwidth requirements for the network by distilling data prior to transfer. A reduction in storage requirements, both on the physical system and for a data pipe, enables more deployment options and strategies. For example, different deployment options and strategies include vehicle mounted (e.g., bus or UPS trucks mapping a geography with cellular backhaul), transportable (e.g., placed in a tower on a limited basis) where ethernet is not available, and man portable (e.g., interactive unit connected to other mobile or fixed units for comparative analysis).

Core capabilities processed on the node at the edge of the network include spectrum reconnaissance, spectrum surveillance with tip and cue, and signal characterization. Spectrum reconnaissance includes automatic capture and production of detail regarding spectrum usage over frequency, geography and time. More actionable information is provided with edge processing, distributed architecture and intelligent data storage. Spectrum surveillance includes automated deconfliction over widebands by comparing real-time data to user supplied, regional and learned data sets and producing alarms. Nodes anSignal characterization provides actionable information. Signals of interest are decoded and demodulated by the system, with location approximation or direction, to improve situational intelligence.

In one embodiment, edge processing of the present invention includes four steps. At step one, first and second derivative FFT analysis is performed in near real time, providing noise floor estimates and signal activity tracking. FIG. 16 is a screenshot illustrating a spectrum with RF signals and related analysis. FIG. 17 is a screenshot illustrating identified signals based on the analysis in FIG. 16. Spectrum in the pink areas and blue areas in FIG. 17 are identified as signals. At step two, analysis is aggregated, signal bandwidths and overall structure are defined, and data is stored to create baselines and be used in reporting. At step three, incoming FFT is compared to existing baselines to find potential conflicts to the baseline. When conflicts are detected, parameters are sent to an event manager (e.g., a logic engine). At step four, the event manager utilizes user supplied knowledge, publicly available data, job manifests and learned information to decide appropriate actions. Action requests such as creating an alarm, sending an e-mail, storing I/Q data, or performing DF are sent to a controller.

A modular approach to system design and distributed computing allows for proper resource management and control when enabled by the right system control solution, which maximizes performance while keeping per-unit cost down. A loosely coupled solution architecture also allows for less costly improvements to the overall network. Parallel processing also enables multiple loosely coupled systems to operate simultaneously without inhibiting each other's independent activities. FIG. 18 is a diagram of a modular architecture according to one embodiment of the present invention. A modular design enables different components to be integrated and updated easily, without the need for costly customization or the never-ending purchase of new equipment, and makes it easier to add in additional hardware/software modules.

Compared to the industry standard tightly coupled architectures increasing complexity and reducing scalability, reliability and security over time, the loosely coupled modular approach provides standardization, consolidation, scalability and governance while reducing cost of operation.

The spectrum monitoring solutions provided in the present invention significantly enhance situational intelligence and physical security, reduces utility complexity and project risk.

The spectrum management systems provided in the present invention are operable to detect and report on incidents in near real time. Remote sensors are placed at site with the capability of capturing and processing RF activity from 40 MHz to 6 GHz. Highly accurate baselines are constructed for automated comparison and conflict detection. Systems are connected to a centralized monitoring and management system, providing alarms with details to a network operations center. On-site systems can also provide messages to additional security systems on-site, such as cameras, to turn them to the appropriate azimuths.

In one embodiment, information such as the presence of a transmission system is operable to be used in an unmanned vehicle recognition system (UVRS) to detect the presence of an unmanned vehicle. The unmanned vehicle is operable to be air-borne, land-based, water-borne, and/or submerged. The detection of certain modulation schemes is operable to be used to identify the presence of mobile phones or mobile radios. This information, coupled with direction finding, provides situational intelligence for informed decision making and rapid response. Measurements and signal intelligence regarding an RF spectrum assist in reducing the risk of financial losses due to theft, vandalism, and power disruptions, providing additional safety for employees and visitors, making other security technologies, such as thermal cameras and IP videos smarter by working in tandem to identify and locate the presence of threats, and capturing and storing I/Q data, which is operable to be utilized as evidence for legal proceedings.

Wireless devices are operable to be utilized across multiple bands. While other monitoring systems are limited on bandwidth (i.e., limited focus) or resolution (making it difficult to see narrowband signals), the systems in the present invention are designed to be more flexible and adaptable and capable of surveying the entire communications environments looking for illicit activity. FIG. 19 illustrates a communications environment according to one embodiment of the present invention.

In one embodiment, a signal characterization engine is configured to provide information including location information and direction, operator name, drone transmission type, and MAC address. All these are actionable information enabling swift resolution. FIG. 20 illustrates an UVRS interface with positive detections, according to one embodiment of the present invention. FIG. 21 lists signal strength measurements according to one embodiment of the present invention.

In one embodiment, the systems of the present invention are operable to be used for mitigating drone threats, identifying and locating jammers, and ensuring communications. The systems of the present invention are designed to identify illicit activity involving use of the electromagnetic spectrum such as drone threats, directed energy/anti-radiation weapons aimed at degrading combat capability (e.g., jammers). The systems of the present invention also bring structure to largely unstructured spectral data enabling clearer communications (interference reduction) and efficient communication mission planning.

Jammers are becoming more prevalent and are operable to be deployed on-site or off premises, making them very difficult to locate. The solutions provided by the present invention automatically send alerts as to the presence of wideband jammers interfering with critical parts of the communications spectrum, and assist in the location of focused jammers which are operable to be very difficult to find. The ability to proactively and rapidly locate jamming devices reduces disruptions in communications, and improves overall security and limits the potential for financial loss. FIG. 22 illustrates a focused jammer in a mobile application according to one embodiment of the present invention. FIG. 23 illustrates a swept RF interference by a jammer according to one embodiment of the present invention.

To maintain security and coordinate operations, consistent and quality communications are imperative. The systems provided in the present invention have multiple deployment strategies and data is operable to be collected and distilled into strength and quality metrics. The data is easy to access in reports. FIG. 24 illustrates data collection, distillation and reporting according to one embodiment of the present invention.

The systems provided in the present invention have the capability of building baselines, detecting when signals exist which are not common for the environment, and creating alerts and automatically starting processes such as direction finding.

The systems provided in the present invention are operable to be used for countering unmanned vehicles, including but not limited to unmanned aerial systems, land-based vehicles, water-borne vehicles, and submerged vehicles. FIG. 25 is a comparison of multiple methodologies for detecting and classifying UAS. Of the methods listed in FIG. 25, RF detection provides the highest level of accuracy in classifying an object as a UAS.

An RF-based counter-UAS system comprises multiple receivers in a single platform. In one embodiment, there are four receivers. Each receiver is operable to scan multiple bands of spectrum looking for UAS signatures. For example, the multiple bands of spectrum include 433 MHz, 900 MHZ, 2.4 GHz, 3.5 GHZ, and 5.8 GHz Base. Each receiver has the capability of scanning a spectrum from 40 MHz to 6 GHz. The receivers are capable of working in tandem for DF applications. Multiple RF-based counter-UAS systems can communicate with each other to extend range of detection and enhance location finding accuracy. The RF-based counter-UAS systems of the present invention comprise proprietary intelligence algorithm on one or multiple GPUs with execution time less than 10 ms. FIG. 26 lists capabilities of an RF-based counter-UAS system according to one embodiment of the present invention. The capabilities of an RF-based counter-UAS system include detection, classification, direction finding, and message creation.

In one embodiment, an RF-based counter-UAS system is operable to be deployed as a long-distance detection model as illustrated in FIG. 27. Four omni-directional antennas are used to create an array for detection and direction finding. In one embodiment, Gimbal-mounted (rotating) defeat antenna is operable to be added. The long-distance detection model is simple to install. In one embodiment, extremely long-distance detection is operable to be obtained with arrays utilizing masts with a height of 8 to 10 meters.

FIG. 28 illustrates features of drones in the OcuSync family. DJI's OcuSync transmission system, utilized by the Mavic Family, is the most advanced commercial UAS System on the market. The drones transmit HD video at 720 and 1080 reliably in environments even with strong radio interference. The drones automatically scan the environment, choosing the frequency band with the lowest interference. The drones are operable to communicate with the controller at nearly 7 km. Links are operable to be established or re-established within one second. Adaptable and automated power settings provide for extending battery life by nearly 100% compared to Lightbridge UASs. FIG. 29 illustrates features of drones in the Lightbridge family. The Phantom Family of UASs utilize the Lightbridge transmission system. The system was developed for long range and robust communication and provides a significant improvement over WiFi UASs, with up to a 5 km communication potential. The system includes 8 selectable channels, which are operable to be altered manually or automatically by the system to avoid interference. A live stream in HDMI or SDI format is sent to the controller for display. The long ranges, adaptability, and ubiquity of OcuSync and Lightbridge systems make them potentially very dangerous. The RF-based counter-UAS systems in the present invention are operable to detect and defeat UASs using these systems.

The RF-based counter-UAS systems in the present invention are operable to detect UASs over a distance of 1.5 kilometers with direction. UASs are operable to be detected and categorized faster than other systems. The RF-based counter-UAS systems can easily integrated into third party systems (e.g., RADAR and camera systems), or act as the common operating platform for other systems for command and control. The RF-based counter-UAS systems are capable for wideband detection from 70 MHz to 6 GHZ, enabling detection of UASs at 433 MHz, 900 MHz, 2.4 GHz, 3.5 GHZ, and 5.8 GHz. The RF-based counter-UAS systems are capable of detecting and direction finding UAS controllers. In one embodiment, unknown and anomalous signals can be categorized as UAS.

In one embodiment, the RF-based counter-UAS systems in the present invention are operable to be used for detecting other unmanned vehicles such as land-based, water-borne, or submerged unmanned vehicles in addition to detecting unmanned aerial vehicles.

In one embodiment, the present invention provides an autonomous and intelligent spectrum monitoring system capable of detecting the presence of wireless activity across extremely wide bands, capturing and performing analysis on highly intermittent signals with short durations automatically, and converting RF data from diverse wireless mobile communication services (e.g., cellular, 2-way, trunked) into knowledge. The autonomous and intelligent spectrum monitoring system of the present invention are advantageous with edge processing, modular architecture, job automation, and distributed sensor network.

Edge processing enables the delivery of a truly autonomous sensor for automated signal recognition and classification and near real-time alarming 24/7, equipped with machine learning algorithms.

A modular architecture increases speed and efficiency, enables more bandwidth to be analyzed (with superior resolution), reduces latency and network traffic (i.e., low backhaul requirements). Logic engines produce relevant alarms, thus limiting false positives.

Job automation allows hardware solutions to be customized to meet operational needs with inclusion of additional receivers and GPUs, cloud or client hosted backend, and third-party integration.

A distributed sensor network supports feature specific applications such as direction finding and drone threat management, capable of LMR and cellular demodulation and assisting prosecution efforts with data storage.

The spectrum monitoring system of the present invention represents a paradigm shift in spectrum management. Edge processing migrates away from the inefficiencies of manual analysis, or the time delays of backhauling large data sets. The spectrum monitoring system of the present invention performs real-time, automated processing at the device level, providing knowledge faster, reducing network traffic and improving application performance with less latency. Modular architecture makes additional development, integration of new features and the incorporation of third party systems easy, and also future-proof capital expenditure. Job automation simplifies operations (e.g., data collection, setting triggers) by enabling the execution of multiple complex tasks, with one click on a user interface. Distributed sensors provide security to critical assets spread across large geographies, linked to a network operations center. Data is operable to be shared to perform location finding and motion tracking.

For critical assets, only certain types of transmitting devices (e.g., radios, phones, sensors) should be present on specified frequencies. The spectrum monitoring system of the present invention learns what is common for a communications environment and creates alarms when an anomalous signal is detected in close proximity. Alerts, along with details such as signal type (e.g., LMR, Mobile, Wi-Fi) and unique characteristics (e.g., radio ID) are posted to a remote interface for further investigation. The spectrum monitoring system of the present invention which is capable of learning, analyzing and creating alarms autonomously provides a heightened level of security for critical assets and infrastructure. FIG. 30 illustrates a spectrum monitoring system detecting an anomalous signal in close proximity of critical infrastructure.

The spectrum monitoring system derives intelligence by collecting, processing, and analyzing spectral environments in near real time. The unique characteristics and signatures of each transmitter are compared automatically to either user supplied or historical data sets. Potential threats are identified quickly and proactively, reducing acts of vandalism, theft and destruction. Advantageously, the spectrum monitoring system of the present invention reduces the risk of financial losses due to theft, vandalism, and power disruptions, provides additional safety for employees and visitors, makes other security technologies including thermal cameras and IP video smarter by working in tandem to identify and locate the presence of threats (with DF functionality), and captures and stores data, which is operable to be utilized as evidence for legal proceedings.

Node devices in the spectrum monitoring system of the present invention are operable to be deployed across large geographies. The spectrum monitoring system is built to interact with third party systems including cameras and big data platforms, providing additional intelligence. All these systems send pre-processed data to a cloud platform and are visualized efficiently on a single interface. FIG. 31 illustrates a system configuration and interface according to one embodiment of the present invention.

Alarms generated at the site are sent to a remote interface, enabling perimeters to be monitored 24/7 from anywhere. Alarm details including transmitter type (e.g., mobile phone), unique identifiers (e.g., radio ID), UAV type, and directions are presented on the interface.

Job automation restructures work flow and the need for configuration management, greatly reducing manual efforts regarding receiver configuration, trigger and alarm management, analytics and reporting, system health management, and conflict and anomalous signal detection.

Not all activity observed in a spectral environment represents a threat. Even in remote locations, LMR radios are operable to be observed. Pedestrians may also be in the area utilizing mobile devices. The spectrum monitoring system of the present invention is equipped with logic to determine the typical makeup of an environment (e.g., common signals based on time of day), proximity, and duration (e.g., time on site). The logic limits false positives to produce alarms that are meaningful. Parameters are operable to be adjusted as required.

FIG. 32 is a screenshot illustrating no alarm going off for an anomalous signal from LMR traffic not in proximity of site according to one embodiment of the present invention.

In one embodiment, the spectrum monitoring system of the present invention enables 24/7 scanning of a local environment, identification of new activities (e.g., LMR, cellular, Wi-Fi), threat assessment capability (e.g., proximity and duration analysis), and alarm creation with details sent via email and posted to a user interface.

In one embodiment, the spectrum monitoring system of the present invention supports a powerful user interface simplifying remote monitoring, greatly improves receiver sensitivity and processing enabling identification of intermittent signals with millisecond durations (e.g., registration events, WhatsApp messaging, background applications), and provides an enhanced logic engine which is operable to identify both signals with long durations (e.g., voice calls, video streaming, data sessions) and repetitive short bursts (e.g., Facebook updates).

In one embodiment, the spectrum monitoring system of the present invention is capable of mobile phone identification from 800-2600 MHZ (covering all mobile activity at site), recognition of intermittent and bursting signals associated with cellular applications, identification of LMR, Wi-Fi, and UAV activity, and determining proximity and limiting false alarms with logic engines.

Node devices in a spectrum monitoring system of the present invention are operable to produce data sets tagged with geographical node location and time. The data sets are operable to be stored on the node devices, or fed to a cloud-based analytics system for historical trend analysis, prediction models, and customer driven deep learning analytics.

Analytics provided by the spectrum monitoring system of the present invention are operable to be used to identify the presence of constant or periodic signals. For example, recognition of the presence of wireless cameras can indicate potential surveillance of a critical asset site. Also for example, the presence of constant or periodic signals can indicate existence of organized groups, attempting to determine normal access patterns for the purpose of espionage or theft.

Analytics provided by the spectrum monitoring system of the present invention can also be used to review patterns before and during an intrusion at several sites and predict next targeted sites.

Analytics provided by the spectrum monitoring system of the present invention can also be used to track contractor and employee visits, both planned and unplanned to the site, to augment data for work flow improvements.

FIG. 33 illustrates a GUI of a remote alarm manager according to one embodiment of the present invention.

FIG. 34 labels different parts of a front panel of a spectrum monitoring device according to one embodiment of the present invention.

FIG. 35 lists all the labels in FIG. 33 representing different part of the front panel of the spectrum monitoring device according to one embodiment of the present invention.

FIG. 36 illustrates a spectrum monitoring device scanning a spectrum from 40 MHz to 6 GHz according to one embodiment of the present invention.

FIG. 37 lists the capabilities of a spectrum monitoring system according to 5 main on-network mobile phone states plus 1 no-network mobile phone state.

A mobile phone in the first main state is active on network, and activities also include short-duration (e.g., milliseconds) activities (e.g., text messages, WhatsApp messages and registration events) besides completing a voice call, engaging in a data session, and streaming video. The first main state lasts 6 to 8 hours typically. Receiver sensitivity for speed and bandwidth and processing are enhanced to enable the capability of intercepting these activities and producing an alarm by the spectrum monitoring system of the present invention.

In the second main state, there are background applications running. To conserve battery life, a mobile phone does not constantly monitor the network, but does “wake up” and check for messages (e.g., every 10 seconds). The mobile phone checks applications including Facebook, SMS, voicemail, email, Twitter, and game challenge notifications. A typical phone sends an update notice (e.g., a request to pull down emails, Facebook messages, etc.) every 90 seconds on average. Background applications such as social media updates are extremely short in duration. To capture these events, receivers in the spectrum monitoring system are doubled (e.g., 2 to 4), the bandwidth of each receiver is doubled (e.g., 40 MHz to 80 MHZ), and software is developed to enhance the system to process the increase in sample (e.g., 10×).

FIG. 38 illustrates a mobile event analysis per one minute intervals according to one embodiment of the present invention.

Events on a mobile phone include background apps (e.g., Facebook, Email, location services, sync apps) with a probability of 90%, active apps (e.g., mobile search, gaming) with a probability of 30%, messaging (e.g., SMS, WhatsApp, Snapchat) with a probability of 15%, voice calls with a probability of 10%. The combined probability gets to 95%.

FIG. 39 is a site cellular survey result according to one embodiment of the present invention. The site cellular survey result reveals there is not active GSM network on site, which means the vast majority of the mobile phones need to be UMTS and LTE capable to have service.

The present invention provides systems, methods and apparatus for automatic signal detection in a radio-frequency (RF) environment. A multiplicity of sensor devices constructed and configured for cross-communication in a nodal network. Each of the multiplicity of sensor devices comprises at least one RF receiver, a generator engine, and an analyzer engine. The at least one RF receiver is configured to measure power levels in the RF environment in real time or near real time and generate fast Fourier transform (FFT) data based on power level data. The generator engine is configured to calculate a power distribution by frequency of the RF environment in real time or near real time, including a first derivative and a second derivative of the FFT data. The analyzer engine is configured to create a baseline based on statistical calculations of the power levels measured in the RF environment for a predetermined period of time, identify at least one conflict situation by comparing the power distribution in real time or near real time to the baseline of the RF environment, and identify at least one signal based on the first derivative and the second derivative of the FFT data in the at least one conflict situation.

Certain modifications and improvements will occur to those skilled in the art upon a reading of the foregoing description. The above-mentioned examples are provided to serve the purpose of clarifying the aspects of the invention and it will be apparent to one skilled in the art that they do not serve to limit the scope of the invention. All modifications and improvements have been deleted herein for the sake of conciseness and readability but are properly within the scope of the present invention.

Claims

1. A system for signal detection or conflict detection in an electromagnetic environment, comprising: at least one electromagnetic receiver configured to measure power levels of activity in the electromagnetic environment; andan analyzer engine;wherein the at least one electromagnetic receiver is configured to generate fast Fourier transform (FFT) data based on the measured power levels in the electromagnetic environment;wherein the analyzer engine is configured to receive the FFT data from the at least one electromagnetic receiver and compare FFT data to at least one baseline to identify at least one conflict or at least one signal in the electromagnetic environment; andwherein the analyzer engine is configured to use at least one machine learning algorithm to learn and update signal detection parameters.

2. The system of claim 1, wherein the analyzer engine is configured to provide location information relating to the at least one signal.

3. The system of claim 1, wherein the system is configured to capture and process data between about 40 MHz to about 6 GHz.

4. The system of claim 1, wherein the analyzer engine is configured to identify at least one signal emitting device.

5. The system of claim 1, wherein the at least one signal comprises intermittent signals and/or bursting signals.

6. The system of claim 1, wherein the at least one signal includes an anomalous signal.

7. The system of claim 1, wherein the analyzer engine is configured to utilize edge processing to analyze the FFT data from the electromagnetic receiver.

8. The system of claim 1, wherein the system is configured to determine an action to take based on the at least one conflict, wherein the action includes creating an alarm, sending an email, storing in-phase and quadrature (I/Q) data, or performing direction finding (DF).

9. The system of claim 1, wherein the analyzer engine is configured to automatically detect the at least one signal by detecting a change in power distribution in the electromagnetic environment.

10. A method for signal detection or conflict detection in an electromagnetic environment, comprising: at least one electromagnetic receiver measuring power levels of activity in the electromagnetic environment;the at least one electromagnetic receiver generating fast Fourier transform (FFT) data based on the measured power levels;an analyzer engine receiving the FFT data and comparing the FFT data to at least one baseline to identify at least one conflict or at least one signal; andthe analyzer engine using at least one machine learning algorithm to learn and update signal detection parameters.

11. The method of claim 10, further comprising providing location information relating to the at least one signal.

12. The method of claim 10, further comprising the analyzer engine identifying at least one signal emitting device.

13. The method of claim 10, wherein the at least one signal includes an anomalous signal.

14. The method of claim 10, further comprising automatically classifying the at least one signal based on the at least one machine learning algorithm.

15. The method of claim 10, further comprising taking an action based on the at least one conflict, wherein the action includes creating an alarm, sending an email, storing in-phase and quadrature (I/Q) data, or performing direction finding (DF).

16. A system for signal detection or conflict detection in an electromagnetic environment, comprising: at least one electromagnetic receiver configured to create measured data based on measurements of the electromagnetic environment; andan analyzer engine;wherein the at least one electromagnetic receiver is configured to generate fast Fourier transform (FFT) data based on the measured data;wherein the analyzer engine is configured to compare FFT data to at least one baseline to identify at least one conflict or at least one signal; andwherein the analyzer engine is configured to use at least one machine learning algorithm to learn and update signal detection parameters or conflict detection parameters based on signal properties.

17. The system of claim 16, wherein the analyzer engine is configured to provide location information relating to the at least one signal.

18. The system of claim 16, wherein the analyzer engine is configured to identify at least one signal emitting device.

19. The system of claim 16, wherein the at least one signal includes an anomalous signal.

20. The system of claim 16, wherein the system is configured to determine an action to take based on the at least one conflict, wherein the action includes creating an alarm, sending an email, storing in-phase and quadrature (I/Q) data, or performing direction finding (DF).