Quantum Neuromorphic Sensor Network

RF-photonic systems present a highly effective alternative to conventional RF receivers, enabling low-loss, long-distance transmission with minimal attenuation and reduced thermal noise—key features for scalable, high-fidelity sensing and quantum networking. Traditional analog processing methods, although well-suited to narrowband applications, are limited in bandwidth and programmability when applied to broadband RF contexts. By transducing RF signals to the optical domain, RF-photonic systems overcome these limitations, preserving signal integrity across extended distances with reduced noise. Nevertheless, RF-photonic transduction encounters significant challenges in maintaining inference accuracy due to shot noise inherent in optical detection. This noise, bounded by the standard quantum limit (SQL), impacts each sample individually, while the Nyquist theorem imposes additional constraints on the sampling rate—challenges that intensify in wideband RF applications with high sampling requirements, restricting RF-photonic sensor's capability to capture essential signal features accurately. To address these issues, we propose an integrated approach that combines a physics-based inner-product engine with an optical squeezing mechanism. The inner-product engine projects the optical signal into a feature space prior to sampling, concentrating critical information within a reduced sample space to enhance resilience against sampling noise. Moreover, the optical squeezer reduces shot noise to below than the SQL, further maximizing the signal-to-noise ratio (SNR). This combined system offers a scalable and efficient RF-photonic transduction solution that significantly improves inference accuracy in complex signal environments, advancing the potential for high-precision RF-photonic applications.