You are here

Machine-Learning Based Sensing and Waveform Adaptation for SDRs Operating in Congested and Contested Environment

Award Information
Agency: Department of Defense
Branch: Army
Contract: W911NF-21-C-0016
Agency Tracking Number: A2-8399
Amount: $1,099,825.00
Phase: Phase II
Program: STTR
Solicitation Topic Code: A19C-T005
Solicitation Number: 19.C
Solicitation Year: 2019
Award Year: 2021
Award Start Date (Proposal Award Date): 2020-12-13
Award End Date (Contract End Date): 2022-06-13
Small Business Information
15378 Avenue of Science Suite 200
San Diego, CA 92128-1111
United States
DUNS: 020817883
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Elettra Venosa
 (858) 332-0700
Business Contact
 David Strobel
Phone: (858) 332-0700
Research Institution
 University of Arizona
 Kirsten Sherman-Haynes
888 N. Euclid Ave Rm 510
Tucson, AZ 88888-9999
United States

 (520) 626-6418
 Nonprofit College or University

The unprecedented growth in wireless demand has caused the radio spectrum to become overcrowded. Modern Software Defined Radios (SDRs) must provide satisfactory performance and agility while operating in congested and contested spectrum environments. To do that, complex learning algorithms must be paired with capable, flexible, and wideband radio hardware. Space Micro and its partner research institution, the University of Arizona, bring together innovations in SDR design and integration of Machine Learning (ML) based software solutions for adapting SDR-based network operation. Together, these innovations will provide the Army with improved radio systems that learn from data, identify patterns, and optimize transmissions in a congested/jammed environment with minimal human intervention. The overarching goal of this STTR project is to design, evaluate, and experimentally demonstrate highly accurate yet computationally efficient machine learning (ML) algorithms for real-time waveform sensing, classification, and channel access over shared spectrum. The targeted spectrum sharing scenarios include both “vertical” sharing (e.g., primary/secondary) as well as “horizontal” sharing, i.e., all friendly wireless systems have equal rights to accessing the spectrum. Coexisting systems exhibit heterogeneity in their access protocols, waveforms, and other transmission parameters, and may operate in the presence of potentially unknown adversarial transmitters, e.g., fake/rogue waveforms. Our use cases cover both commercial and military applications. Phase 1 effort explored various ML-based sensing, classification, spectrum-usage prediction, and channel access algorithms, considering vertical sharing for some tasks and horizontal sharing for others. Due to the short duration of Phase 1 (3 months), we did not fully explore the design space, and focused on providing preliminary assessment via Matlab simulations. We also did not incorporate rogue transmissions in our classification designs and focused primarily on competing but legitimate waveforms. No over-the-air experiments were conducted. For Phase 2, we plan to further investigate and enhance the most promising classifiers of Phase 1; expand the scope of our effort to include classification under rogue waveforms; provide comprehensive simulation-based evaluation; implement several of our algorithms on an SDR platform, and test the efficacy of these algorithms through SDR-based experimentation. Our Phase 2 efforts will culminate in a comprehensive demo, presented to the Army.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government