Delay-Doppler Domain Pulse Design for OTFS-NOMA
Michel Kulhandjian, Hovannes Kulhandjian, Gunes Karabulut Kurt, Halim Yanikomeroglu
TL;DR
The paper tackles the challenge of achieving high spectral efficiency in OTFS-based NOMA under high mobility by introducing Hermite-function delay-Doppler pulses to realize per-user orthogonality in the delay-Doppler domain. It extends the sufficient (bi)orthogonality train-pulse concept by replacing square-root Nyquist pulses with Hermite functions, leveraging their orthogonality and time-frequency localization. Key findings show that Hermite-based OTFS-NOMA outperforms traditional OTFS-NOMA variants (PDM-NOMA and CDM-NOMA) in BER on high-mobility channels, while maintaining low complexity dominated by OTFS demodulation; its spectral efficiency scales with the NOMA spreading length $K$, achieving a $K$-fold gain over OTFS-CDM-NOMA. The work provides a practical pathway to improved BER and spectral efficiency in OTFS-NOMA for 6G+ scenarios with challenging channel dynamics, using orthogonal DD-domain pulses to facilitate user separation without heavy computational loads.
Abstract
We address the challenge of developing an orthogonal time-frequency space (OTFS)-based non-orthogonal multiple access (NOMA) system where each user is modulated using orthogonal pulses in the delay Doppler domain. Building upon the concept of the sufficient (bi)orthogonality train-pulse [1], we extend this idea by introducing Hermite functions, known for their orthogonality properties. Simulation results demonstrate that our proposed Hermite functions outperform the traditional OTFS-NOMA schemes, including power-domain (PDM) NOMA and code-domain (CDM) NOMA, in terms of bit error rate (BER) over a high-mobility channel. The algorithm's complexity is minimal, primarily involving the demodulation of OTFS. The spectrum efficiency of Hermite-based OTFS-NOMA is K times that of OTFS-CDM-NOMA scheme, where K is the spreading length of the NOMA waveform.
