A preprint reports that a proposed OTFS synchronization method produced timing-offset error below one sample across all tested signal-to-noise ratios in numerical simulations of a high-mobility link. The same study reported that its synchronization and joint-estimation approach incurred less than a 2 dB loss in bit-error performance compared with a perfect-channel-state-information case.
The result comes from a modeled single-user OTFS system, with the receiver tested against simulated RF impairments and an extended vehicular A channel model. It is therefore an evaluation of algorithms in a defined computer model, rather than a measurement from a physical radio.
A receiver built around three estimates
The paper asks whether three proposed elements can estimate and compensate for OTFS RF impairments under high mobility: a pilot-aided timing-offset method, a joint estimator for carrier-frequency offset and the effective channel, and a detector that works in the delay-Doppler domain.
The first stage uses a low-peak-to-average-power-ratio PCP pilot and exploits the pilot's dual periodicity to estimate the timing offset. In ordinary terms, this is the part of the pipeline that measures how far the received frame is shifted from its expected timing.
The next stage uses a maximum-likelihood technique to estimate carrier-frequency offset and the effective channel jointly. It represents that channel with a complex-exponential basis-expansion model known as CE-BEM, allowing the calculation to account for the combined estimation problem rather than treating the two quantities separately.
The final step formulates a linear detection model in delay-Doppler coordinates to address interference left by the RF impairments. Together, the three stages form the proposed synchronization, estimation and detection pipeline evaluated in the simulations.
The evaluation was numerical rather than a participant study: its test objects were simulated OTFS frames and channel-and-impairment configurations. The analysis compared a configuration with 32 delay bins with one using 64 delay bins.
Timing and frequency estimates moved in different directions
The timing result was reported as a mean absolute error, or MAE. The maximum-peak timing-offset estimator's error stayed below one sample at every tested signal-to-noise ratio, and the authors described that amount of error as absorbable by the cyclic prefix, or CP. Exact MAE values for each signal-to-noise ratio were not tabulated in the supplied analysis.
The delay-bin comparison also showed a trade-off. Timing-offset accuracy was reported to deteriorate as the number of delay bins increased. The supplied results give no numerical effect size or uncertainty interval for that change.
The pattern for the other estimates was different. The proposed carrier-frequency-offset estimator was reported to have mean squared error, or MSE, in the order of 10−4. The paper also reported more accurate carrier-frequency and effective-channel estimates when the number of delay bins increased, without stating a numerical effect size or uncertainty interval for that improvement.
For data recovery, the proposed synchronization and joint-estimation techniques were reported to incur less than a 2 dB loss compared with the perfect-CSI case. Here, bit-error rate, or BER, is the performance measure used for the recovered data, while perfect CSI supplies an idealized channel-information reference within the simulation.
The test conditions set the boundary of the result
The simulated frames used a 1.0417-microsecond sampling period, a 5.9-gigahertz carrier frequency, either 32 or 64 delay bins, and 16 Doppler bins per frame. The channel was the extended vehicular A, or EVA, model with a 2.7-kilohertz Doppler spread.
The simulated phase noise followed a first-order autoregressive model with β3dB set to 400 hertz. Amplitude imbalance was varied uniformly from 0 to 2 dB, phase imbalance from 0 to 5 degrees, timing offset from 0 to M−1, and carrier-frequency offset over (−0.5, 0.5]. The basis-expansion settings were Q = 7 and K = 2.
These choices mean that the reported numbers describe the stated frame, channel and impairment configuration. They do not establish how the pipeline would perform under conditions outside those simulated settings. The supplied analysis also reports no replicate counts, confidence intervals or formal sample-size rationale.
The uncertainty is also unevenly reported across the results. The study gives the order of magnitude for the carrier-frequency-offset MSE and directional changes for the delay-bin comparisons, but not full numerical curves, interval estimates or exact BER values in the supplied text.
A simulation result, not a hardware verdict
The authors interpret the estimation and BER analyses as demonstrating suitability for practical low-cost radios in high-mobility scenarios. That interpretation remains bounded by the evidence described here: the document evaluates a modeled single-user system through numerical simulation and provides no physical-hardware result.
The study therefore answers a narrower engineering question. It reports how the proposed pipeline behaved under the selected OTFS, EVA-channel and RF-impairment settings; it does not establish performance on real low-cost radio hardware or outside the stated simulation configuration.
The document is an arXiv preprint identified as arXiv:2608.20257v1 and dated 20 August 2026.
The work was conducted within the MULTIPLY-6G project, funded through the Smart Networks and Services Joint Undertaking under Horizon Europe, and supported by Research Ireland through the US-Ireland R&D Partnership Programme.
Paper data and sources
Original title: Synchronization and Channel Estimation of OTFS with RF Impairments
Authors: Sanoopkumar P. S., Mohsen Bayat, Stephen McWade, Arman Farhang
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text