Adaptive AI-Driven Waveform Design Service

The issuing agency is seeking an adaptive artificial intelligence-driven waveform design service to autonomously select, synthesize, and optimize transmit waveforms on the fly. The agency needs to solve the technical challenge of bridging the gap between high-level machine learning algorithms, specifically deep reinforcement learning, and low-latency, real-time hardware execution. The required solution must enable neural network inference to execute within micro-seconds to alter parameters on a pulse-to-pulse or coherent processing interval basis. The scope of work requires optimizing models for deployment on edge-computing architectures, such as software-defined radios backed by field-programmable gate arrays or specialized system-on-chip processors. The project covers ensuring high performance of these models without exceeding strict size, weight, and power constraints. Performance of the work will take place offsite. Eligibility is strictly limited to onshore organizations based in the United States. The agency has not disclosed a specific budget and is currently looking for proposals to determine the cost. No key dates are specified in the sourcing materials.