OTFS downlink synchronization for 6G NTN

Non-Terrestrial Networks (NTN), especially based on Low Earth Orbits (LEO), represent a key enabler of sixth-generation (6G) wireless networks. Support for satellite links in 5G was introduced in 3GPP Release 17, retaining CP-OFDM and DFT-s-OFDM to minimize the impact on the already standardized waveform for the terrestrial component. With the recently initiated definition of 6G, the investigation of new waveforms has begun to address the objective of natively integrating NTN characteristics. This has led to the consideration of waveforms that are more resilient to typical NTN channel propagation impairments, including high Doppler shifts and large differential delays. Among the different candidates, OTFS and its variants have emerged as promising alternatives due to their inherent robustness against significant Doppler shifts, which are particularly characteristic of LEO constellations. Since any new waveform, including OTFS, necessitates the comprehensive analysis and development of receiver algorithms for synchronization, estimation, and detection, in this paper, we propose a downlink synchronization procedure for OTFS featuring an enhanced timing offset estimation algorithm and a detection scheme for the received signal, also in multipath scenarios. The synchronization signal consists of a pilot symbol from a QPSK constellation and a Zadoff-Chu (ZC) sequence. Specifically, we use the pilot symbol for the detection of the incoming signal, identification of the presence of multiple paths, and the estimation of the timing offset (TO) in the delay-time (DT) domain. The estimation of the Doppler shift is performed in two steps: first, the fractional carrier frequency offset (FCFO) is estimated by means of the pilot symbol in the DT domain, then the estimation of the integer carrier frequency offset (ICFO) is conducted through the ZC sequence in the delay-Doppler domain. The numerical results show that the proposed timing offset estimation algorithm outperforms conventional two-steps algorithm, especially for medium-high SNR values. In terms of Doppler shift estimation, the ZC sequence demonstrates greater reliability than the maximum length sequence (MLS) under low-SNR conditions.