Enhanced DFT-Based Channel Estimator for Leakage Effect Mitigation in OFDM Systems
Abstract: The leakage effect occurs in a DFT-based channel estimator (CE) due to the presence of guard bands, where the energy of the channel impulse response (CIR) is dispersed to other estimated taps. Furthermore, some of the CIR energy leaks outside of the estimation range, and fails to be captured in the channel estimation. This letter presents an enhanced DFT-based CE for OFDM systems. The proposed method is composed of two techniques (spectral extrapolation and circular shift) to mitigate the leakage effect, followed by a denoising post-processing to enhance the estimation accuracy. Through performance analysis, it is demonstrated that the spectral extrapolation eliminates the leakage effect corresponding to the first CIR tap, and the circular shift improves estimation accuracy by incorporating the leaked CIR energy in the channel estimation. Simulation results show that the proposed method has superior performance, compared with conventional CEs. In addition, the proposed method is feasible from a computational complexity perspective. Existing system: Many studies on channel estimation in OFDM systems, however, often neglect the guard bands. Recently, there have been studies on compressed sensing based
channel estimation. However, most of the work did not take the GB into account, when they were devised for OFDM systems. To solve performance degradation from the leakage effect, other approaches have been proposed. In, a leakage suppression estimator was proposed, which turned out to be equivalent to the linear minimum mean square error (LMMSE) CE. The regularization-based method in performs time domain post-processing on the CIR estimate for the purpose of leakage nulling. The method in can be interpreted as a variant of the LMMSE CE. Accordingly, these methods require huge complexity by the matrix inversion. Proposed system: One of the major drawbacks of OFDM systems is high out-of-band emission. To prevent interference to the adjacent spectrum, some portions of band-edge subcarriers are reserved for a guard band (GB) in practical OFDM systems. However, the null subcarriers within the GB destroy the orthogonality of the DFT matrix. Subsequently, the DFT matrix is ill conditioned, and this leads to distortion of the DFT-based CIR estimation, which is the so-called leakage effect. Recently, OFDM is being applied to Internet of Thing systems, such as IEEE 802.11ah and 3GPP NB-IoT, where the importance of low complexity is far more crucial. Therefore, studies for an accurate and complexity-efficient CE that can mitigate the leakage effect by the GB is still of key importance for OFDM-based communication systems. Advantages: Orthogonal frequency division multiplexing (OFDM) has been widely applied in wireless communication systems due to its high bandwidth efficiency. To support high order modulation for high data rate transmission, accurate channel estimation is indispensable in OFDM systems. To meet the requirements of both high accuracy and low complexity, discrete Fourier transform (DFT)-based channel estimators (CEs) have been widely used. In the DFTbased CE, the channel impulse response (CIR) is estimated by exploiting the orthonormality of of the DFT matrix. One of the major drawbacks of OFDM systems is high out-of-band emission. To prevent interference to the adjacent spectrum, some portions of band-edge subcarriers are reserved for a guard
band (GB) in practical OFDM systems. However, the null subcarriers within the GB destroy the orthogonality of the DFT matrix. Disadvantages: Many studies on channel estimation in OFDM systems, however, often neglect the guard bands. Recently, there have been studies on compressed sensing based channel estimation. However, most of the work did not take the GB into account, when they were devised for OFDM systems. To solve performance degradation from the leakage effect, other approaches have been proposed. In, a leakage suppression estimator was proposed, which turned out to be equivalent to the linear minimum mean square error (LMMSE) CE. The regularization-based method in performs time domain post-processing on the CIR estimate for the purpose of leakage nulling. Modules: Orthogonal frequency division multiplexing (OFDM): Has been widely applied in wireless communication systems due to its high bandwidth efficiency. To support high order modulation for high data rate transmission, accurate channel estimation is indispensable in OFDM systems. To meet the requirements of both high accuracy and low complexity, discrete Fourier transform (DFT)-based channel estimators (CEs) have been widely used. In the DFTbased CE, the channel impulse response (CIR) is estimated by exploiting the orthonormality of the DFT matrix. One of the major drawbacks of OFDM systems is high out-of-band emission. To prevent interference to the adjacent spectrum, some portions of band-edge subcarriers are reserved for a guard band (GB) in practical OFDM systems. However, the null subcarriers within the GB destroy the orthogonality of the DFT matrix. Subsequently, the DFT matrix is ill conditioned, and this leads to distortion of the DFT-based CIR estimation, which is the so-called leakage effect. Guard band: Many studies on channel estimation in OFDM systems, however, often neglect the guard bands. Recently, there have been studies on compressed sensing based
channel estimation. However, most of the work did not take the GB into account, when they were devised for OFDM systems. To solve performance degradation from the leakage effect, other approaches have been proposed. In, a leakage suppression estimator was proposed, which turned out to be equivalent to the linear minimum mean square error (LMMSE) CE. The regularization-based method in performs time domain post-processing on the CIR estimate for the purpose of leakage nulling. The method in can be interpreted as a variant of the LMMSE CE. Accordingly, these methods require huge complexity by the matrix inversion. In this letter, an enhanced DFT-based CE is proposed to effectively mitigate the leakage effect. The proposed method is composed of spectral extrapolation, circular shift and denoising techniques. Selection of in circular shift: The optimal in the circular shift can be obtained by minimizing eleakage. However, this requires knowledge of the delay and variance for each tap. Practically, such information may not be available in advance. To select a proper n the circular shift, a uniform channel is considered as suggested for robust CE design. shows eleakage as a function of _ for a uniform channel with different lengths, Lh. It is observed that eleakage exhibits an error floor around the minimum, irrespective of the channel length, and these error floors overlap. Thus, it is proposed to use a predetermined for the various channel environments. This would be a practical (suboptimal) solution from an implementation perspective.