Researches And DSP Implementation Of Channel Estimation For OFDM System In High Speed Mobile Environment

Due to the characteristics involving high frequency spectrum efficiency, powerful anti-multipath effect ability, fast transmission rate and simple engineering implementation etc., orthogonal frequency division multiplexing (OFDM) has been widely applied to telecommunications and now becomes one of the key technologies in long term evolution (LTE) and LTE-Advanced(LTE-A) systems. Latest decades witnessed dramatic developments in mobile communication area. In spite of the fact that conventional voice service is still in most needed, data service like multimedia now also accounts for big market sharing and tends to increase remarkably year by year. Meanwhile, with the large nationwide construction in high-speed railway, conventional communication services like texting and phone calling cannot meet the market demands any more. So the new mobile communication demands in high-speed scenario also drive the technology evolution in telecommunication. In high-speed mobility scenario, the communication channel is fast varied, so accurate channel estimation is one of the key factors affecting the performance of OFDM system. Aiming at the simulation of fast channel estimation of OFDM system in high-mobility environment, the thesis focuses on its DSP implementation on the software radio platform.In this thesis, the fading characteristics of mobile communications are first introduced, followed by the description of multipath fading and Doppler frequency shift in small-scale fading. Multipath delay will cause some degree of inter-symbol interference (ISI), the Doppler shift will cause a rapid channel change in the symbol period. To deal with this situation, it is quite essential to perform the equilibrium and demodulation in receive side to estimate the channel accurately. Then analyze the fundamental principle and system structure of OFDM system.Next, conventional methods of channel estimation are introduced. As for the channel estimation in pilot carriers, the algorithms including least squares (LS), linear minimum mean square error (LMMSE), DFT-based algorithm and singular value decomposition (SVD) based algorithm are commonly used. By contrast, in terms of the channel estimation in non-pilot carriers, the interpolation methods are usually used, among which linear interpolation and second order interpolation are introduced in detail in this thesis. While conventional channel estimation methods are usually suitable for quasi-static channel, considering the fast-varying channel environment, we apply the model of Basis Expansion Model (BEM) here to track the channel variation. This thesis researches several typical implementation algorithms of BEM coefficients. Then, the BEM channel estimation algorithms are analyzed and the BEM channel estimation in time domain is finally implemented in DSP in this thesis.In the final section, the implementation diagram of channel estimation through DSP is introduced. In this section, the DSP implementation key functional components are introduced, particularly the design methods and optimization process of each key functional module in the implementation process of channel estimation. Through the comparison between the results of hardware test and simulation, the feasibility of implementing channel estimation in DSP is verified.