End-to-End OTFS Prototyping on Low-Cost SDRs With Energy-Retentive Message Passing Detection

Abstract

The next generation of wireless systems includes the integration of new communication paradigms such as low-earth-orbit satellites, and intelligent sensing and communications, which operate in highly dynamic channels. For the 4th and 5th generation systems, the dominant waveform that has been adopted is the orthogonal frequency division multiplexing (OFDM). However, OFDM performance suffers in such dynamic environments with high Doppler spread. The orthogonal time-frequency space (OTFS) has emerged as a promising waveform for addressing these issues. In this paper, we compare OTFS and OFDM under modeled doubly dispersive channels and analyze three channel-matrix sparsification strategies for message-passing-based OTFS symbol detection. The proposed framework is evaluated through Monte Carlo simulations and an SDR-based channel-emulation experiment. In the latter, a software-generated doubly dispersive channel is applied to the digital transmit samples before they are passed through two stationary PlutoSDRs separated by approximately 1 m. The resulting measurements determine whether the detector trends observed in simulation persist after the waveform traverses the SDR conversion and indoor RF chain, including synchronization, uncompensated oscillator offset and phase noise, and analog-front-end effects. Because the Doppler components are introduced numerically rather than through physical motion or RF-path variation, these measurements do not constitute a true high-mobility over-the-air trial. Under the controlled emulated conditions, OTFS achieves lower Bit Error Rate (BER) than the OFDM baseline, while the proposed energy-retentive sparsification provides a practical complexity–accuracy tradeoff for OTFS receiver design.