Research On Channel Equalization Techniques For O-OFDM System In Visible Light Communication Based On Machine Learning

With the widespread advancement of new-generation information technologies such as cloud computing,big data,artificial intelligence and 5G,people have higher demands for communication speed and capacity.However,traditional radio frequency(RF)communication technology is limited by spectrum resources and cannot meet the growing data transmission needs.Visible light communication(VLC)technology,as an emerging wireless communication technology,has attracted wide attention from researchers.VLC technology not only has abundant spectrum resources,but also has advantages such as high transmission rate,green energy saving,and anti-electromagnetic interference.It has a broad application prospect.However,indoor VLC technology still faces many technical challenges to be popularized in practical applications.Among them,the impact of VLC channel on system performance is an urgent problem to be solved.Optical signals experience degradation in system performance due to various factors such as background noise,inter-symbol interference(ISI),and multipath fading caused by reflections from walls and objects during propagation through the channel.It is important to address these issues to ensure effective signal transmission.In view of this,this paper analyzes the impact of interference factors such as path loss,multipath effect,background noise on system performance in VLC channels.And for this problem,this paper studies the machine learning(ML)equalization method widely used at present.Firstly,aiming at the shortcomings of traditional K-means equalization method,a channel equalization method based on improved K-means algorithm is proposed.In addition,this paper also considers the impact of VLC channel on system performance when the receiver is in a mobile state and proposes a channel compensation method based on deep learning(DL).The specific research contents are as follows:1.A channel equalization method for optical orthogonal frequency division multiplexing(O-OFDM)system based on improved K-means algorithm is proposed.The traditional Kmeans algorithm is a clustering algorithm based on iterative updating of center points.Its equalization technique requires constant iteration to achieve the optimal clustering result,so it has a high complexity.The paper proposes the Improvement Center K-means(IC-Kmeans)equalization method,which incorporates a training sequence to determine clustering centers without repeated iterations.This approach reduces algorithm complexity while maintaining a consistent bit error rate(BER)performance similar to K-means equalization.In addition,Kmeans equalization method is difficult to accurately distinguish mixed data on the boundary between clusters and clusters,so the equalization effect is not ideal.This paper proposes the Neural Network(NN)Based Improvement Center K-means(NNIC-Kmeans)equalization method.First,it uses a neural network to map data to three-dimensional space to enlarge the distance between different constellation cluster interlaced signals.Then it uses IC-Kmeans algorithm to cluster to achieve better equalization effect.Through Monte Carlo simulations,this study compares the complexity and BER performance between the proposed method and traditional equalization methods.The results indicate that the IC-Kmeans equalization method effectively reduces complexity while maintaining BER performance,as the required number of iterations for equalization is reduced to 1.Additionally,the NNIC-Kmeans equalization method significantly improves BER performance.2.Considering the signal reception situation when the receiver is in a mobile state and proposing a channel compensation method for indoor visible light communication based on deep learning.This paper aims at the problem of channel characteristics changing with time when the receiver is in a mobile state in indoor VLC systems.It adopts two typical mobile path models to analyze the impact of interference factors such as noise and multipath effect in the channel on system BER performance during mobile process and proposes a channel compensation method based on deep learning.This method uses long short-term memory algorithm(LSTM)to equalize received signals to offset interference factors caused by channel during mobile process.Through Monte Carlo simulations,this study analyzes the performance improvement of the proposed method on BER under different signal-to-noise ratio conditions in two mobile path models.The results indicate that the LSTM equalization method proposed in this paper significantly enhances the system’s BER performance at various signal-to-noise ratios.