New Method Study On Peak-to-Average Ratio Suppression And Channel Estimation And Equalization In Multicarrier Systems

Multicarrier modulation technology is widely accepted by modern wireless communication systems because of its strong capability to resist multipath fading.Especially the orthogonal frequency division multiplexing(OFDM)has become the standard waveform for many contemporary communication systems.However,OFDM cannot meet the needs of all future communication application scenarios,and the filter bank multicarrier(FBMC)system has characteristics such as low spectral sidelobes,high spectral efficiency,flexible parameter configuration and low sensitivity to time synchronization.The characteristics described above can well compensate for the shortcomings of OFDM systems,so FBMC is considered to be an important complementary waveform to OFDM in future wireless communications.The high peak-to-average power ratio(PAPR)of the transmitted signal and the estimation and equalization of the doubly selective channel are the two major challenges for OFDM and FBMC systems.From the perspective of improving PAPR suppression performance and reducing computational complexity,effective PAPR suppression methods for OFDM and FBMC signals are proposed respectively based on the in-depth analysis of existing methods.In this dissertation,the channel estimation and equalization problem in the multicarrier system is constructed as a deep learning task,and the neural networks model-based channel estimation and equalization scheme is proposed.The main works and contributions of this dissertation are listed as follows:1.Aiming at the problem of high PAPR in the OFDM system,the artificial bee colony(ABC)algorithm based SLM and PTS schemes are studied.The traditional ABC-SLM and ABC-PTS schemes have the disadvantage of low efficiency of phase factor search.Based on the in-depth analysis of the reasons for the mismatch between ABC algorithm and SLM and PTS schemes,SLM and PTS schemes based on discreteartificial bee colony algorithm(DABC),which work directly in discrete space and have high search efficiency,are proposed.Simulation results show that under the same computational complexity,the proposed DABC-SLM and DABC-PTS schemes have better PAPR suppression performance than that of other SLM and PTS schemes based on group intelligence optimization algorithms.2.Aiming at the problem of high PAPR in the FBMC/OQAM system,a conversion vector-based low-complexity SLM scheme is studied.The traditional SLM scheme has high computational complexity due to the need for multiple IFFT operations,so is impractical in low power applications such as machine-type communication(MTC).A conversion vector based low-complexity dispersive SLM(C-DSLM)scheme is proposed in this dissertation.In the C-DSLM scheme,the candidate signals are generated by multiplying the original signal by the cyclic shift of the conversion vectors with only a few non-zero elements,avoiding a lot of high complexity IFFT calculation.The complexity evaluation and simulation results show that the proposed C-DSLM scheme can obtain similar PAPR suppression performance when the computational complexity is much lower than that of the DSLM scheme.3.Aiming at the problem of estimation and equalization of doubly selective channels in multicarrier systems,deep learning(DL)based solutions are studied.Traditional channel estimation and equalization methods are highly dependent on tractable mathematical channel models,which are generally assumed to be linearly stationary and follow a Gaussian statistical distribution.However,actual wireless communication systems have many defects that cannot be captured with an accurate mathematical model or are affected by unknown factors.Especially in the doubly selective channels,traditional methods tend to suffer from performance degradation.To jump out of the limitations of the traditional analytical methods,the channel estimation and equalization problem of the multicarrier system is constructed as a DL task from the perspective of learning.For the OFDM system,a Res DNN-based channel estimation and equalization(Res DNN-CE)scheme is proposed.In the Res DNN-CE scheme,the traditional channel estimation and equalization module and demapping module are replaced by the Res DNN model.Simulation results show that the channel estimation and equalization performance of the Res DNN-CE scheme are significantly better thanthat of the traditional analytical method and DL-CE scheme under the influence of unfavorable factors such as insufficient pilot number,no CP,and clipping noise.For the FBMC system,a deep neural network(DNN)based channel estimation and equalization(DNN-CE)scheme is proposed,and good performance is obtained.To further improve the estimation and equalization performance of time-varying channels of the DNN-CE scheme,based on the in-depth study of the internal correlation between FBMC signals and Bidirectional Long short-term memory(BLSTM)networks,a deep BLSTM model based channel estimation and equalization scheme is proposed.