On Optimal Turbo Decoding of Wideband MIMO-OFDM Systems Under Imperfect Channel State Information

Pablo Piantanida, Sajad Sadough, Pierre DuhamelarXiv 2007 · arXiv

TL;DR

This paper tackles the challenge of decoding wireless signals in MIMO-OFDM systems when the receiver only has an imperfect estimate of the communication channel. They introduce a novel, robust decoding metric that significantly improves performance compared to standard methods, especially when channel estimation is poor, without adding computational complexity.

What Problem Does This Paper Address?

Modern wireless communication systems, like those used in Ultra-Wide-Band (UWB) and for high-speed data, increasingly rely on sophisticated techniques such as Multiband Orthogonal Frequency Division Multiplexing (MB-OFDM) and Multiple-Input Multiple-Output (MIMO) antenna arrays. These technologies aim to increase data rates and spectral efficiency. A common scenario in these systems is that the wireless channel between the transmitter and receiver changes over time. To decode the signal, the receiver needs to know the characteristics of this channel. However, obtaining a perfect channel estimate is difficult and often relies on transmitting pilot symbols, which are limited in number. This leads to an "imperfect channel state information" (CSIR) at the receiver.

The core problem is that standard decoding algorithms, which assume perfect knowledge of the channel, perform poorly when faced with these estimation errors. A common workaround, called "mismatched ML decoding," simply uses the imperfect estimate as if it were perfect. While widely used for performance evaluation, this approach is suboptimal. The paper asks: how can we design a practical encoder/decoder that achieves the best performance even with significant channel estimation errors?

Methods

The authors propose a new decoding metric derived within a Bayesian framework. Instead of directly using the imperfect channel estimate, they average the likelihood function over all possible true channel states, conditioned on the estimated channel. This averaging is performed using the posterior probability density function of the true channel given its estimate, which captures the uncertainty introduced by the estimation process.

This "improved ML detection" metric is then applied to the demapping stage of Bit Interleaved Coded Modulation (BICM) receivers in both MB-OFDM and MIMO-OFDM systems. The paper details how this metric is formulated and used in the context of iterative decoding, where the receiver refines its estimate of the transmitted bits over multiple iterations. Crucially, this new metric is designed to be computationally similar to the mismatched approach, meaning it doesn't introduce significant additional complexity.

Datasets

Not discussed in the paper. The evaluation is based on simulated channel models.

Key Insights and Ideas

  • Robustness through Averaging: The core idea of averaging the likelihood over channel uncertainty is a powerful way to mitigate the negative impact of imperfect channel estimation. It acknowledges the inherent uncertainty rather than treating the estimate as ground truth.
  • Bayesian Framework for Practicality: Using a Bayesian approach to derive the metric allows for a principled way to incorporate the statistics of the channel estimation error, leading to a more accurate and robust decoding strategy.
  • No Added Complexity: A significant achievement is that the proposed improved metric is computationally comparable to the standard mismatched approach, making it practical for real-world implementation without demanding more processing power.
  • Performance Gains Across Scenarios: The improved metric shows substantial gains in bit error rate (BER) across various channel conditions (realistic UWB and Rayleigh fading) and for different numbers of pilot symbols used for estimation.

Key Results

The paper presents simulation results demonstrating the effectiveness of the proposed decoder.

  • Significant BER Improvement: In scenarios with limited pilot symbols (e.g., N=1 or N=2), the proposed "modified ML" decoder significantly outperforms the "mismatched ML" decoder. For instance, in a realistic UWB channel (CM1) with 16-QAM and Gray labeling, using only N=1 pilot symbol, the modified decoder performs nearly as well as the mismatched decoder using N=2 pilots.
  • Reduced SNR Requirement: For a target BER of 10⁻³, the modified decoder requires approximately 1.5 dB less SNR compared to the mismatched decoder when using theoretical Rayleigh fading channels.
  • Impact of Pilot Count: The performance gap between the proposed and mismatched decoders narrows as the number of pilot symbols increases. This is because with more pilots, the channel estimate becomes more accurate, and the proposed metric converges towards the classical Euclidean distance metric.
  • MIMO Performance: For a 2x2 MIMO-OFDM system, the improved decoder shows a notable SNR advantage (around 1.5 dB for a BER of 10⁻⁵) when using only N=2 pilots compared to the mismatched decoder.

Limitations and Future Work

  • Limitations: The paper focuses on specific system models (MB-OFDM, MIMO-OFDM) and assumes a time-invariant channel over a frame. The performance gains are demonstrated through simulations using specific channel models and modulation schemes (16-QAM).
  • Future Work: The authors suggest that this work provides a foundation for further research into robust decoding strategies under imperfect channel knowledge. Potential future directions could include exploring adaptive pilot design, investigating the impact of different interleaving schemes, or extending the approach to more complex channel models and higher-order modulation schemes. The paper also hints at the possibility of using this metric to derive achievable outage rates, which could be a valuable area for future study.

Summary updated

BibTeX
@misc{piantanida2007optimal,
  author = {Pablo Piantanida and Sajad Sadough and Pierre Duhamel},
  title = {{On Optimal Turbo Decoding of Wideband MIMO-OFDM Systems Under Imperfect Channel State Information}},
  year = {2007},
  eprint = {0708.1413},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/0708.1413},
}

Loading related content...

Figure