A low-PAPR Pilot Design and Optimization for OTFS Modulation

TL;DR

The paper tackles the challenge of achieving high spectral efficiency (SE) in OTFS modulation while keeping the peak-to-average power ratio (PAPR) low in doubly-selective channels. It introduces a data-pilot superimposed design (S1D) based on Chu sequences, optimized energy split between pilot and data, and an iterative channel estimation/equalization procedure that leverages sparse recovery via OMP in the delay-Doppler domain. Compared to embedded pilots (EP), the S1D approach significantly improves SE (about a 57% gain) and reduces PAPR (by roughly 3 dB), albeit with some NMSE/BER trade-offs that can be mitigated through iterative processing and the choice of equalizer. The method is particularly advantageous for tight power budgets (e.g., satellite IoT) where higher SE and lower PAPR are essential, and it offers a flexible framework for balancing accuracy, interference, and complexity.

Abstract

Orthogonal time frequency space (OTFS) modulation has been proposed recently as a new waveform in the context of doubly-selective multi-path channels. This article proposes a novel pilot design that improves OTFS spectral efficiency (SE) while reducing its peak-to-average power ratio (PAPR). Instead of adopting an embedded data-orthogonal pilot for channel estimation, our scheme relies on Chu sequences superimposed to data symbols. We optimize the construction by investigating the best energy split between pilot and data symbols. Two equalizers, and an iterative channel estimation and equalization procedure are considered. We present extensive numerical results of relevant performance metrics, including the normalized mean squared error of the estimator, bit error rate, PAPR and SE. Our results show that, while the embedded pilot scheme estimates the channel more accurately, our approach yields a better tradeoff by achieving much higher spectral efficiency and lower PAPR.

Figures (8)

• Figure 1: IDZT-OTFS communication pipeline.
• Figure 2: Our proposed S1D data-pilot design (left) vs the embedded pilot construction from EP (right). The shaded elements represent data frame locations that would not affect the channel estimation procedure but are kept empty to enable the use of simpler equalizers.
• Figure 3: NMSE of channel estimation as a function of the pilot energy.
• Figure 4: BER as a function of the pilot energy.
• Figure 5: Average spectral efficiency as a function of the pilot energy.