| In the troposcatter communication system,long-distance signal transmission can be realized by using tropospheric scattering to transmit signals.However,long transmission distance of general scattering communication system will cause serious path loss,and the scatterers in the troposphere will have multi-path effect on the signal.Affected by the atmosphere,the troposphere is extremely unstable,and the scattering communication channel has serious time-frequency selective fading.In order to achieve reliable communication,it is necessary to adopt a multi-carrier communication system with anti-frequency selective fading characteristics.SC-FDMA transmission system is just a good choice to obtain multi-carrier waveform characteristics with low peak-to-average ratio.In the SC-FDMA transmission system,an interleaved mapping method adopted in SC-IFDMA make the subcarriers interwoven and far apart,thus reducing the impact of deep fading in frequency domain.In this way,a certain frequency domain diversity gain can be obtained.Therefore,this thesis conducts technical research based on SC-IFDMA system.Due to the complex environment of the troposcatter communication channel,the receiver thus has higher requirements for signal equalization technology.Among the existing equalization technologies,nonlinear equalization technology can resist intersymbol interference and frequency selective fading with good equalization characteristics.Therefore,this thesis studies the nonlinear equalization technology of SC-IFDMA system.The channel estimation error will have a crucial impact on the effect of equalization.The scattering communication system has the characteristic of low signal-to-noise ratio,while the traditional channel estimation algorithm has significant defects under the low signal-tonoise ratio.Poor channel estimation performance will lead to poor decision feedback equalization performance.In view of excellent nonlinear function mapping ability of neural network,this thesis introduces neural network to improve the performance of channel estimation and integrates it with traditional noise prediction decision feedback to propose a neural network-assisted decision feedback equalization architecture.The traditional Fourier transform can realize the conversion between the time domain and the frequency domain.As a generalized form of the Fourier transform,the fractional Fourier transform has the characteristic of adjustable order and has stronger flexibility compared with traditional Fourier transform.By selecting different orders to make it match with the channel characteristics better,more excellent performance than the traditional frequency domain equalization can be obtained.So,this paper explores the definition and properties of fractional Fourier transform,and proposes a kind of decision feedback equalization technology based on fractional Fourier transform.The simulation results show that the performance of this algorithm is significantly improved compared with the existing algorithms.In view of the problem that the proposed new method requires iteration,the complexity is slightly higher,this paper proposes a low-complexity decision feedback equalization technology based on fractional Fourier transform.The simulation results show that the performance of this low-complexity decision feedback equalization technology based on fractional Fourier transform is similar to that of the decision feedback equalization technology based on fractional Fourier transform proposed in this paper,but the computational complexity is reduced significantly.This paper is of great significance to the research and application of nonlinear equalization technology in SC-IFDMA system. |
