Direct Reconstruction of Saturated Samples in Band-Limited OFDM Signals

Kyong Hwan Jin, Gain Kim, Yusuf Leblebici, Jong Chul Ye, Michael UnserarXiv 2018 · OpenReview

Abstract

Given a set of samples, a few of them being possibly saturated, we propose an efficient algorithm in order to cancel saturation while reconstructing band-limited signals. Our method satisfies a minimum-loss constraint and relies on sinc-related bases. It involves matrix inversion and is a direct, non-iterative approach. It consists of two main steps: (i) regression, to estimate the expansion coefficients of the signal model; (ii) interpolation, to restore an estimated value for those samples that are saturated. Because the proposed method is free from tuning parameters, it is hardware-friendly and we expect that it will be particularly useful in the context of orthogonal frequency-division multiplexing (OFDM). There, the high peak-to-average power ratio of the transmitted signal results in a challenging decoding stage in the presence of saturation, which causes significant decoding errors due to the nonlinearity of amplifiers and receivers, ultimately resulting in band distortion and information loss. Our experiments on realistic simulations confirm that our proposed reconstruction of the saturated samples can significantly reduce transmission errors in modern high-throughput digital-communication receivers.

Introduction: The Challenge of Signal Saturation

Orthogonal Frequency-Division Multiplexing (OFDM) serves as the foundation for modern wireless and wireline communication standards, including 5G and high-speed broadband. A significant drawback of OFDM is its high Peak-to-Average Power Ratio (PAPR). When multiple subcarriers align constructively, they generate voltage spikes that exceed the linear operating range of power amplifiers or Analog-to-Digital Converters (ADCs).

Conceptual overview of signal saturation and the proposed reconstruction of true values from clipped samples.

This phenomenon, known as "clipping" or saturation, results in in-band distortion (data errors) and out-of-band radiation (interference). Traditional solutions often rely on complex iterative algorithms or require additional "side channels" to transmit correction data. This research introduces a direct, non-iterative method to recover these lost peaks using only the available unsaturated samples.

Methodology: Direct, Non-Iterative Reconstruction

The proposed algorithm treats signal recovery as a mathematical problem within Reproducing-Kernel Hilbert Spaces (RKHS), leveraging the band-limited nature of OFDM signals. The reconstruction process is divided into two primary stages:

  1. Regression: The system utilizes a set of reliable, unsaturated samples to perform a matrix inversion, determining the underlying expansion coefficients of the signal model.
  2. Interpolation: These coefficients are then used to "fill in" the missing data for saturated samples through straightforward matrix-vector multiplication.
Integration of the non-iterative reconstruction block within a standard OFDM transmitter and receiver pipeline.

To ensure the method is viable for real-time hardware implementation (such as ASICs or FPGAs), the researchers employed a Sliding Window approach. By processing a small local neighborhood of samples around each saturation event rather than the entire data stream, matrix sizes remain small (e.g., $8 \times 8$), significantly reducing computational overhead.

Key Insights and Results

The primary advantage of this approach is its "hardware-friendly" nature. Unlike prior compressive sensing or Bayesian methods that require multiple iterations to converge, this one-shot matrix solution is computationally efficient. Furthermore, the method ensures sampling consistency, meaning the reconstructed signal perfectly aligns with the known, unsaturated data points.

Key performance metrics from realistic numerical simulations include:

  • Bit-Error Rate (BER) Reduction: In ideal wireless scenarios, the BER was reduced from an unusable 0.14 to near zero after reconstruction.
  • Robustness: The algorithm maintains performance gains even in the presence of additive white Gaussian noise (AWGN), outperforming standard clipping-only receivers by several orders of magnitude.
  • Efficiency: The computational cost is dominated by a simple sorting step and a small matrix inversion, which can be optimized using pre-calculated lookup tables for common saturation patterns.

Conclusion and Future Outlook

While the algorithm is highly effective, its performance can degrade if saturation density is too high—specifically, if there are insufficient "good" samples in a local neighborhood to perform reliable regression. Additionally, it assumes knowledge of the hardware's saturation thresholds, which are typically fixed ADC specifications.

Future work points toward direct silicon implementation. Integrating this non-iterative reconstruction block into ASICs could allow it to operate at GHz speeds, potentially becoming a standard front-end component for next-generation high-throughput digital receivers.

Summary updated

Abstract

Given a set of samples, a few of them being possibly saturated, we propose an efficient algorithm in order to cancel saturation while reconstructing band-limited signals. Our method satisfies a minimum-loss constraint and relies on sinc-related bases. It involves matrix inversion and is a direct, non-iterative approach. It consists of two main steps: (i) regression, to estimate the expansion coefficients of the signal model; (ii) interpolation, to restore an estimated value for those samples that are saturated. Because the proposed method is free from tuning parameters, it is hardware-friendly and we expect that it will be particularly useful in the context of orthogonal frequency-division multiplexing. There, the high peak-to-average power ratio of the transmitted signal results in a challenging decoding stage in the presence of saturation, which causes significant decoding errors due to the nonlinearity of amplifiers and receivers, ultimately resulting in band distortion and information loss. Our experiments on realistic simulations confirm that our proposed reconstruction of the saturated samples can significantly reduce transmission errors in modern high-throughput digital-communication receivers.
BibTeX
@misc{jin2018direct,
  author = {Kyong Hwan Jin and Gain Kim and Yusuf Leblebici and Jong Chul Ye and Michael Unser},
  title = {{Direct Reconstruction of Saturated Samples in Band-Limited OFDM Signals}},
  year = {2018},
  eprint = {1809.07188},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1809.07188},
}
Figure