ML-based LPD Chaotic Communications

Low Probability of Detection (LPD) is a critical characteristic in modern covert waveform design to protect Soldiers and enable mission success. Increasingly, adversaries are able to detect LPD waveforms designed using traditional methods such as Direct Sequence Spread-Spectrum by using cyclostationary analysis techniques. In this Phase I effort, Cryptic Vector will develop a prototype featureless LPD waveform physical layer based on coherent Chaotic Sequence Spread-Spectrum and model cyclostationary attack methods to validate the waveform’s LPD properties. Multi-user access is achieved using a Code Division Multiple Access scheme extended for chaotic sequences. Emerging research in using machine learning techniques for chaotic sequence synchronization will be leveraged to overcome the challenges in contested, noisy, and GPS-denied radio frequency (RF) environments that have previously prevented wider adoption of chaotic communications for LPD applications. A Mobile Ad Hoc Network (MANET) architecture will be investigated, based on the open standard IPv6 over Low-Power Wireless Personal Area Networks (6LoWPAN) adaptation layer providing easy integration of PLI, chat, data, and voice services. The 6LoWPAN network scales to large deployments providing a decentralized, fully (single hop) or partially (multi-hop) connected mesh with no single point of failure and no additional infrastructure. Low-cost hardware solutions will be analyzed to provide sufficient computational and RF performance, taking advantage of Cryptic Vector’s experience developing COTS-based proof-of-concept platforms. Finally, a basic demonstration of the prototype physical layer will be provided using COTS software-defined radios at the end of the Phase I base period, with an integrated file transfer demonstration at the end of the Phase I option period.