研究目的
To present a novel time-domain equalizer for visible light communication (VLC) system using machine learning (ML) method, specifically employing discrete Fourier transform spread orthogonal frequency division multiplexing (DFT-S-OFDM) as modem scheme and convolution neural network (CNN) as kernel processing unit of equalizer.
研究成果
The CNN-based time-domain equalizer significantly improves BER performance of DFT-S-OFDM systems under channels with relatively small 3-dB bandwidth, outperforming traditional methods like RLS and OMP. Future work includes investigating more optimized network structures.
研究不足
The complexity of the proposed CNN-based equalizer is higher than traditional methods, requiring significant computational resources and storage space. However, the method can be accelerated by GPU or FPGA for real-time performance.
1：Experimental Design and Method Selection:
The study employs DFT-S-OFDM as modem scheme and CNN as the kernel processing unit of the equalizer. The methodology includes estimating channel state information (CSI) from training sequence and recovering transmitted symbols based on the estimated CSI.
2：Sample Selection and Data Sources:
The transmitted data is generated in pseudorandom binary sequence (PRBS) form.
3：List of Experimental Equipment and Materials:
The neural network is based on Python 3.5.4 with TensorFlow 1.3.0. RLS is accomplished using functions provided by MATLAB R2017a.
4：4 with TensorFlow RLS is accomplished using functions provided by MATLAB R2017a.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: The proposed equalizer processes the received signal through a CNN and fully connected layers to recover the transmitted symbols. The training employs stochastic gradient descent (SDG) framework with a mini-batch of the training-set.
5：Data Analysis Methods:
The performance is evaluated based on bit error rate (BER) improvement and comparison with traditional methods like RLS and OMP.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
