Abstract
In Orthogonal Frequency-Division Multiple Access (OFDMA) systems, pilot signals often cover only a partial frequency band. This partial allocation causes significant signal "leakage" across time-domain taps when using standard Discrete Fourier Transform (DFT)-based channel estimation, resulting in a persistent Mean Square Error (MSE) floor. This paper presents a symmetric extension technique that mirrors the partial frequency response before processing. This approach effectively cancels leakage through symmetry, significantly improving estimation accuracy with minimal computational overhead.
Introduction
Channel estimation in OFDM systems typically employs the DFT-based method to suppress noise. By transforming the frequency-domain channel estimate into the time domain via an Inverse DFT (IDFT), noise-only taps can be zeroed out before transforming back to the frequency domain. However, in OFDMA, users are assigned partial subcarrier sets. The sharp rectangular windowing of this partial band in the frequency domain causes the signal to spread or "leak" across the entire time-domain window.
Conventional DFT methods cannot distinguish this leakage from noise. Consequently, filtering noise also discards signal power, leading to an MSE floor that does not diminish even as Signal-to-Noise Ratio (SNR) increases. This paper addresses this limitation by introducing a continuity-enhancing symmetric extension.
Methodology
The proposed method smooths the edges of the partial frequency response to concentrate signal energy in the time domain. The process follows these steps:
- Initial Estimation: Obtain the $M$-point partial frequency response using Least Squares (LS) estimation.
- Symmetric Extension: Mirror the $M$ points to create a $2M$-length sequence, ensuring a smoother transition at the band edges.
- 2M-point IDFT: Transform the extended sequence to the time domain. The mirrored structure facilitates self-cancellation of leakage terms.
- Time-Domain Filtering: Apply a threshold ($L_{symmetric}$) to eliminate noise-dominated taps.
- Reconstruction: Perform a $2M$-point DFT and combine the results to extract the final $M$-point channel estimate.
This architecture is computationally efficient, requiring only a doubling of the FFT size, which is negligible for modern hardware.
Results and Discussion
The performance of the symmetric extension method was validated through computer simulations of a 512-subcarrier OFDMA system over a six-path Rayleigh fading channel.
Key metrics included the Leakage-to-Noise Ratio (LNR) and MSE. As shown in the comparative analysis, the symmetric method reduces LNR by several orders of magnitude compared to conventional DFT methods. This reduction allows the estimator to "break" the MSE floor. At high SNR (e.g., 40 dB), the symmetric extension method achieves an MSE near $10^{-4}$, whereas the conventional method plateaus at approximately $10^{-2}$.
These improvements in channel estimation translate directly to enhanced Bit Error Rate (BER) performance, particularly in high-SNR regimes where conventional methods typically fail to improve.
Conclusion
Symmetric extension effectively mitigates time-domain leakage in partial-band OFDMA channel estimation. By creating periodic-like continuity in the frequency domain, the method concentrates signal energy and eliminates the persistent error floor of traditional DFT-based estimators. Future research may explore adaptive thresholding for the time-domain filter and the application of this technique to high-mobility MIMO-OFDMA systems.
Summary updated