Jeff Zurita · Senior Application Engineer
Deepwave Digital and Spectrum Management
At the recent 2021 GNU Radio Conference, Dr. John Chapin, Special Advisor for Spectrum at the National Science Foundation, gave a keynote presentation (shown below) on the “Future Interference Management and Spectrum Monitoring.” In his presentation, Dr. Chapin outlined the new ways technology could enable more effective forms of spectrum management and the need that these novel techniques will have for spectrum monitoring to obtain real time data about spectrum usage.
Data from spectrum monitoring sensors would be used to support policy decisions as well as build trust in the RF spectrum user community that they will be protected from interference. The sensors and data collected will have to meet high standards for trust and privacy, in addition to accuracy of measurement.
General requirements for spectrum sensors and the management and control of a sensor network have been captured in an IEEE standard, IEEE 802.15.22–3, the IEEE Standard for Spectrum Characterization and Occupancy Sensing. The IEEE standard describes the spectrum sensor as an “entity that provides distributed RF sensing through a remotely accessible API”.
The IEEE standard was created anticipating a network of spectrum sensors with heterogeneous capabilities and likely from different manufacturers. The sensors therefore will be expected to provide structured responses to a set of standard commands and provide the results of measurements in a standardized way.
At Deepwave Digital, we support Dr. Chapin’s vision of interference mitigation by widespread spectrum sensing. We believe that a software defined radio (SDR) will be required for these capabilities to be fully realized. Deepwave offers our product, the Artificial Intelligence radio Transceiver (AIR-T) SDR as the platform upon which to develop a reference spectrum characterization and occupancy (SCOS) sensor as defined in the IEEE standard.
Key Enablers
The capabilities of the AIR-T make it an ideal platform on which to develop and deploy the spectrum sensor as described by the IEEE Standard due to the following features:
- Bandwidth: The AIR-T is tunable from 30 MHz to 6 GHz with up to a 100 MHz usable bandwidth (125 MSPS).
- Compute Resources: The AIR-T has FPGA, CPU, and GPU processing elements. The ample processing resources provides the AIR-T as a spectrum sensor the capability to compute, storage, and sensing resources without the need for intervention by the sensor manager. The GPU in particular makes the AIR-T capable of executing deep learning operations, such as automatic signal identification at the deployed edge location without the need for reach-back to a data center.
- Interfaces: The AIR-T includes standard ethernet interface, USB, and M.2 SATA internal storage. These interfaces allow command and control access as well as the delivery of sensed spectrum data to the spectrum management system over a variety of transport protocols.
- Ruggedized, Edge Deployment: The AIR-T model AIR8201 was designed for an outdoors, exposed environment, as would be expected for edge deployment. The AIR-T includes ruggedized components and a case for ingress protection, certification pending.
- Operating System and Trusted APIs: The AIR-T has its own Linux based operating system, AirStack, allowing for remote administration using industry standard protocols. The included software supports spectrum sensing and deep learning applications written in any of most common, industry standard programming languages.
- Virtualization: The AIR-T supports virtual environments, allowing spectrum sensor applications to be uploaded and exchanged by Spectrum Manager and Data Client entities as changing conditions require.
Spectrum Management
The AIR-T is activity used in a spectrum management application as the spectrum sensor for the Citizen’s Band Radio Service (CBRS), a shared spectrum band. Deepwave Digital’s AIR-T signal identification technology is being deployed commercially in a 5G wireless system as a fully developed and tested product, with a maturity equivalent to TRL 7 or higher. The signal identification technology was developed to enable shared use of the spectrum between commercial and government users as part of the rollout of the 3.5 GHz frequency band as part of the new 5G system. In CBRS, the 5G service provider is required to monitor spectrum usage to determine if a priority user, a US Navy radar system, is operating.
Deepwave Digital developed the deep learning signal classifier to detect the priority user, resulting in an FCC certified signal classification solution that combines traditional GPU-accelerated DSP techniques with advanced deep learning methods. The accuracy of the AIR-T as a spectrum classifier was a key factor in obtaining the FCC certification, leading to the ongoing deployment of AIR-T based spectrum sensors across the CBRS network.
The performance and deployment of the Deepwave Digital AIR-T in the CBRS Environment Sensing Capability network and the capabilities that the AIR-T brings to the spectrum research community, make the AIR-T the platform of choice upon which to develop a reference IEEE SCOS Sensor.
For more information, please contact us here.