Frame Detection And Channel Estimation For DCO-OFDM Lifi Systems

The available radio spectrum has become scarce,and visible light communication(VLC)which utilizes 400-800 THz for data transmission has effectively solved this problem.In order to make light flash quickly and achieve high data rates.An Light Emitting Diode(LED)and a photodiode are usually utilized as transmitter and receiver,respectively.In order to meet the requirements of high data rates and real positive signals,Direct Current Biased Optical-Orthogonal Frequency Division Multiplexing(DCO-OFDM)is widely used for visible light communication systems,and can effectively solve the effect of inter-symbol interference caused by an LED.DCO-OFDM systems are sensitive to timing errors and require synchronization,i.e.,frame detection,to achieve accurate demodulation of the received signals.However,the LED limited bandwidth has a huge impact on classic correlation-based time-domain frame detection methods,resulting in their inapplicability to DCO-OFDM VLC systems with the LED limited bandwidth considered.On the one hand,since intensity modulation and direct detection(IM/DD)is used,there is no carrier frequency offset(CFO)for VLC systems.On the other hand,there is no inter-symbol interference between signals in frequency domain when correct frame detection is performed.Therefore,this thesis proposes the idea of frequency-domain frame detection.Specifically,this thesis proposes single-subcarrier based frame detection(SCFD)and multi-subcarrier based frame detection(MCFD)methods that reduce the influence of the LED limited bandwidth on frame detection.SCFD is to use the slope of difference of received frequency-domain signals on the same subcarrier over multiple DCO-OFDM blocks to conduct frame detection.MCFD is to use the difference of received frequency-domain signals on the different subcarriers over multiple DCO-OFDM blocks.Simulation results show that the two proposed methods not only perform better than classic time-domain methods in terms of probability of false detection,but also outperforms on the complexity and spectral efficiency.Channel estimation is also an important process to ensure accurate demodulation.The VLC channels are determinate,different from indeterminate radio frequency channels.They are only related to a number of parameters such as the light radiance angle,the light incidence angle,etc.Therefore,this thesis transforms the estimation of the visible light channel into the estimation of some parameters.Two channel estimation methods are proposed: linear fitting(LF)and minimum of imaginary part(MIC)based methods.LF method utilizes the approximate linearity of the modulus of channel frequency response.The LEDlimited bandwidth is estimated before calculating the line-of-sight(Lo S)component of channel.MIC method utilizes that the Lo S component is real,and this method calculates the imaginary component of product of actual channel and the channel generated by the LED limited bandwidth.The LED limited bandwidth is determined when the imaginary component reaches its minimum.Then,Lo S component is calculated.The simulation results show that the channel estimation methods not only outperform the traditional least square method in terms of mean square error,but also behave better on complexity.Moreover,the proposed frame detection and channel estimation methods can be conducted with the same set of pilots,reducing the training overhead.Compared to some conventional methods,the proposed schemes also behave better on bit error rate(BER).