Abstract: Methods, systems, and apparatus for learned encoding and decoding in wireless networks are disclosed. A method includes receiving one or more radio frequency (RF) signals that encode communications data and do not include pilot data. The RF signals are processed to obtain a representation of the encoded communications data, which is provided to a machine learning model. A decoded version of the encoded communications data is obtained as an output of the machine learning model without reliance on pilot data. A reconstruction of the communications data is then generated based on the decoded version. The techniques facilitate pilot-free transmission modes, such as a zero-demodulation reference signal (DMRS) mode, within existing slot structures to reduce spectral overhead and increase capacity. The machine learning model can be configured using learned constellations or modulation codebooks to perform blind channel estimation and equalization.

Abstract: A method for wireless communication using a service side-channel signaling and authentication at the physical level. This method comprising the steps of creating at least one transmitting node and one receiving node within a wireless communication channel. Then choosing primary message and a secondary message and generating a valid transmission tag. Then superimposing the valid transmission tag and creating a set of secret codebooks in order to form a side-channel. Then applying key equivocation metric to measure key information leakage to an eavesdropper. Then transmitting said primary message and said secondary message. Then a receiver receiving said primary and said secondary messages detecting fingerprint estimating data combining a key set with the estimated data sending data and key set to a matrix to generate a secret codebook searching for valid tags authenticating and recovering side information.