Auxiliary Factor Method to Remove ISI of Nyquist Filters

Zijian Zhou, Lifeng Lin, Bingli JiaoarXiv 2022 · DOI · arXiv

TL;DR

This paper introduces the Auxiliary Factor (AF) Method, a novel approach to completely eliminate Intersymbol Interference (ISI) in practical communication systems. By adding a calculated compensation vector to the transmit signal block, the authors achieve perfect zero-ISI performance regardless of the filter's length, while simultaneously increasing robustness against timing jitter.

What Problem Does This Paper Address?

In digital communications, the "Nyquist first condition" provides a theoretical framework for transmitting signals without intersymbol interference (ISI). However, a gap exists between theory and practice.

The ideal Nyquist filter requires an infinite impulse response. In the real world, engineers must use Finite Impulse Response (FIR) filters, which truncate the signal using a "window." This truncation prevents the system from ever reaching a true zero-ISI state. Traditionally, the only way to reduce this residual ISI was to increase the filter's length (more "taps"), which adds complexity and latency but still never achieves perfection. The authors aim to bridge this gap by finding a way to reach zero-ISI without needing infinitely long filters.

Methods

The core innovation is the Auxiliary Factor (AF) Method. Instead of trying to design a "perfect" filter, the authors modify the input signal itself.

  1. Signal Compensation: They introduce an auxiliary factor vector $\mathbf{z}$ that is added to the transmit signal block $\mathbf{s}$, creating a compensated signal $\mathbf{s'} = \mathbf{s} + \mathbf{z}$.
  2. Closed-Form Solution: The authors derive a mathematical solution to find the exact values of $\mathbf{z}$ that will result in zero ISI at the receiver.
  3. Matrix Decomposition: To make the calculation efficient, they represent the filtering process as a matrix operation. They decompose the system matrix into a banded Toeplitz matrix (representing the standard filter action) and specific sub-matrices that account for "head" and "terminal" distortions (the boundary effects of the FIR window).
  4. Efficient Inversion: By using the Trench algorithm and circular decomposition, they provide a way to calculate these compensation factors with a complexity of $O((K+1)\log_2(K+1))$, where $K$ is the block length.

This method is designed to work with "paired filters"—one at the transmitter for bandwidth limiting and a matched filter at the receiver for maximizing the signal-to-noise ratio (SNR).

Datasets

As a theoretical and algorithmic paper in signal processing, it does not use external "big data" datasets. Instead, it evaluates the method through rigorous simulations using:

  • Standard Pulse Shapes: Square-Root Raised Cosine (SRRC) and "Better Than Raised Cosine" (BTRC) filters.
  • Modulation Schemes: BPSK and 16-QAM.
  • Channel Model: Additive White Gaussian Noise (AWGN).
  • Parameters: Various roll-off factors ($\beta$) and filter orders ($M$).

Key Insights and Ideas

  • Length Independence: The most significant "aha" moment is that zero-ISI is achieved regardless of the window length. You can use a relatively short, efficient filter and still get the performance of an "ideal" infinite filter.
  • Neutralizing Boundary Effects: Most FIR studies struggle with the "head" and "terminal" parts of a signal block where the filter isn't "full" of data. This paper explicitly models these boundary distortions as matrix components and cancels them out.
  • Jitter Robustness: Beyond just fixing ISI at the perfect sampling point, the AF method makes the entire pulse shape more "stable," significantly reducing Bit Error Rate (BER) when the receiver's timing is slightly off (timing jitter).
  • Minimal Power Overhead: The method only slightly increases the Peak-to-Average Power Ratio (PAPR)—less than 6% for BPSK—meaning it doesn't require significantly more expensive power amplifiers.

Key Results

  • Perfect Zero-ISI: In simulations, the relative Root-Mean-Square (RMS) error for the AF-based methods was zero across all tested filter orders and roll-off factors, whereas conventional filters always showed residual error.
  • BER Performance: Under timing jitter conditions, the AF-SRRC and AF-BTRC methods maintained significantly lower Bit Error Rates than their conventional counterparts.
  • Ideal Matching: The BER performance of the AF methods matched the "ideal" theoretical results (as if no filters were used at all) because the ISI was completely neutralized.

Limitations and Future Work

  • Transmission Delay: The primary drawback is latency. Because the AFs are calculated based on a signal block, the transmitter must wait to process the entire block before sending. This makes the current method less suitable for ultra-low-latency applications like real-time gaming or certain industrial controls.
  • Computational Complexity: While the authors optimized the math, it still requires more processing power than simple standard filtering.
  • Extension to Other Filters: While the paper focuses on paired Nyquist filters, the authors suggest the method could be applied generally to any Nyquist filter setup, which remains a path for future verification.

Summary updated

Abstract

As has been known, the Nyquist first condition promises no intersymbol interference (ISI) as derived in the frequency domain. However, the practical implementation using the FIR filter truncates the Fourier transform by its window and prevents the mathematical calculation from reaching the ideal solution at zero-ISI. For obtaining better results, an increase in the window's length is required in general. To address this problem, a new approach is presented by using auxiliary factors (AFs) to compensate shortcomings of the truncated Fourier transform and remove the ISI completely, regardless of the window's length. In addition, the performance in the presence of the timing jitter is also improved significantly. The closed-form solution of the AFs is derived and the effectiveness is confirmed by the simulation results. Finally, the problems of the transmission delay and additional calculation complexity are analysed.
BibTeX
@misc{zhou2022auxiliary,
  author = {Zijian Zhou and Lifeng Lin and Bingli Jiao},
  title = {{Auxiliary Factor Method to Remove ISI of Nyquist Filters}},
  year = {2022},
  eprint = {2209.01739},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2209.01739},
}
Figure