BER Analysis of Decision-Feedback Multiple Symbol Detection in Noncoherent MIMO Ultra-Wideband Systems

Taotao Wang, Tiejun Lv, Hui Gao, Yueming Lu

TL;DR

This paper introduces a novel Decision-Feedback Multiple Symbol Detection (DF-MSD) scheme for noncoherent MIMO Ultra-Wideband (UWB) systems using Differential Space-Time Block Codes (DSTBC). The DF-MSD significantly reduces the computational complexity of existing Multiple Symbol Detection (MSD) methods while achieving comparable or even improved bit error rate (BER) performance, making advanced UWB communication systems more practical.

What Problem Does This Paper Address?

Ultra-Wideband (UWB) technology offers high data rates and precise localization, making it attractive for various applications like wireless sensor networks and personal area networks. When combined with Multiple-Input Multiple-Output (MIMO) antenna arrays, UWB systems can achieve spatial diversity gains. Differential Space-Time Block Codes (DSTBC) are particularly useful for noncoherent MIMO UWB systems, as they enable diversity gain without requiring channel state information (CSI), which is notoriously difficult to estimate in UWB channels.

However, traditional symbol-by-symbol differential detection (DD) in these systems suffers from performance degradation compared to coherent detection. While Multiple Symbol Detection (MSD) offers a significant performance boost by considering multiple symbols jointly, its computational complexity grows exponentially with the observation window size, rendering it impractical for high-performance UWB systems. This paper addresses the need for a high-performance detection scheme that overcomes the complexity limitations of MSD in noncoherent MIMO UWB systems.

Methods

The core contribution of this paper is the proposal of a Decision-Feedback Multiple Symbol Detection (DF-MSD) scheme. The DF-MSD builds upon the MSD framework but incorporates a decision-feedback mechanism. Instead of performing an exhaustive search over all possible symbol combinations within the observation window, the DF-MSD uses previously detected symbols to "clean up" the current waveform template. This effectively reduces the search space for the current symbol detection.

The authors derive the theoretical bit error rate (BER) for a genie-aided DF-MSD, where the feedback symbols are assumed to be perfectly correct. This provides a theoretical lower bound for the real DF-MSD. The paper also analyzes the signal and noise components within the detection metric and derives the conditional probability density functions of the relevant random variables, enabling the calculation of the theoretical BER.

Datasets

Not discussed in the paper. The authors use simulated channel models based on the IEEE 802.15.3a CM2 model for their experiments.

Key Insights and Ideas

  • Complexity-Performance Trade-off: The paper demonstrates that by intelligently feeding back previous decisions, the significant computational burden of MSD can be substantially reduced without a proportional loss in performance.
  • Practicality of DF-MSD: The DF-MSD scheme offers a more practical approach to achieving high performance in noncoherent MIMO UWB systems, bridging the gap between theoretical gains and real-world implementation feasibility.
  • Genie-Aided Analysis as a Lower Bound: The theoretical analysis of the genie-aided DF-MSD provides a valuable benchmark and a realistic lower bound for the performance of the actual DF-MSD receiver, which is crucial for system design and evaluation.
  • Error Propagation Impact: The paper acknowledges that the real DF-MSD may slightly underperform the genie-aided version due to error propagation from feedback symbols, especially at low signal-to-noise ratios (Eb/No).

Key Results

  • The proposed DF-MSD scheme achieves a solid performance improvement over the standard symbol-by-symbol DD scheme.
  • The DF-MSD offers comparable BER performance to the full MSD scheme but with significantly reduced computational complexity.
  • The theoretical BER analysis for the genie-aided DF-MSD provides a lower bound that closely matches simulated results, validating the theoretical framework.
  • For larger observation window sizes (M=20), the DF-MSD performance approaches that of ideal Rake reception for Alamuti space-time coded MIMO-UWB systems, with a gap of only 3 dB at a BER of 10⁻⁶.

Limitations and Future Work

  • Approximation in Noise Analysis: The paper notes that the approximation of the noise component A₃ in equation (17) leads to a slight discrepancy between the theoretical and simulated BER for the genie-aided DF-MSD.
  • Error Propagation: The real DF-MSD receiver's performance is slightly degraded compared to the genie-aided version due to error propagation from feedback symbols, particularly at low Eb/No.
  • Future Work: The authors suggest that the theoretical lower bound derived for the genie-aided DF-MSD can be used to guide the design of soft detectors in future work. The DF-MSD scheme's ability to improve performance with increasing observation window size without prohibitive complexity suggests its potential for even higher performance gains with larger M values.

Summary updated

BibTeX
@article{wang2013ber,
  author = {Taotao Wang and Tiejun Lv and Hui Gao and Yueming Lu},
  title = {{BER Analysis of Decision-Feedback Multiple Symbol Detection in Noncoherent MIMO Ultra-Wideband Systems}},
  journal = {IEEE Trans. Vehicular Technology},
  year = {2013},
  url = {https://openreview.net/forum?id=1amsiaTyp5},
}
Figure