An Adaptive Deep Learning Architecture with FPGA Acceleration for Continuously Monitoring and Characterizing
Abstract
The goal of the proposal is to continuously monitor and characterize operations and signals in congested, contested, and constrained electromagnetic spectrum (EMS) environments for communications and radars, therefore to better meet DoD s command, control, and communication needs on their battlefields and beyond. Due to the technology evolution as well as 5G and beyond which provides high capacity, faster speeds, world-wide connectivity, terrestrial and non-terrestrial communications, it has been harder for the warfighters to have freedom of action within the EMS to be successfully operational in congested, contested and constrained EMS environments globally. To ensure that the U.S. military maintains their ability to operate in EMS and retrieve the “freedom of maneuver” in future by dynamically accessing EMS securely and reliably, we propose to (1) design an incremental deep learning architecture along with region proposals which will be scalable and adaptable to estimate the regions of all active signals, incorporate new operations, scenarios, and signals through incrementally learning new operations in an online manner and updating the network without interruption and network retraining; (2) characterize the operations including normal communications, unintentional interference from adjacent band operations, and intentional interference like jammers, classify the modulation types and waveforms of the signals including 4G Long-Term Evolution (LTE), 5G New Radio (NR), and radar waveforms as well as locate the operations and signals in time, frequency, and spatial domains; (3) reconfigure radio hardware and software to either coexist with other operations or interfere and block malicious operations; and (4) demonstrate the efficacy of the proposed algorithms using commercial off-theshelf wideband mixed signal front-end and field programmable gate array (FPGA). The deliverables will be the algorithms and software, and a low-cost and flexible proof-of-concept prototype. Both graduate and undergraduate students will be trained through this research project
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 18, 2025
- Source ID
- N001742210008
Entities
People
- Ruolin Zhou
Organizations
- United States Navy
- University of Massachusetts Dartmouth