Research Of Fast Time Varying Channel Estimation Based On BEM

In the 21^st Century,the numerous intelligent terminals are emerging,and the traffic system is developing with the tendency of high-speed and densification,which demand that the new generation of mobile communication system not only supports the high rate data transmission,but also ensures the reliable communication in the high-speed mobile environments.These higher requirements would bring several new challenges on the next generation of bandwidth-limited communication systems.For catering the demand of the reliable and high rate transmission under the high-velocity mobility,the long term evolution?LTE?standard of the new generation of communication systems adopts the single-carrier frequency division multiple access?SC-FDMA?and orthogonal frequency division multiplexing access?OFDMA?techniques.Compared with previous generations,the 4G can effectively improve the spectrum efficiency and the system capacity,decrease the inter-symbol interference and eliminate the effect of multipath delay and Doppler shift during the data transmission phases.However,in the high speed mobile environment,the great Doppler shift would cause the time-selective fading,which destroys the orthogonality among the sub-carriers and causes the inter-carrier interferences.The challenges above-mentioned will greatly increase the difficulty of related channel estimation techniques.This paper focuses on the channel estimation techniques of SC-FDMA/OFDMA systems in high speed mobile environments,the key achievements are listed as follows:1)Concentrating on the fast time-varying channel estimation techniques of the SC-FDMA system,we propose two novel low-complexity basis expansion model?BEM?-based channel estimation algorithms in view of the low estimation precision of the conventional interpolation methods or the high computational complexity of the BEM algorithm.The advantages include:a)the proposed simplified time-frequency-BEM algorithm utilizes the channel property of time-frequency domain to simplify the formulation of transferring the channel time domain response to frequency domain.b)The proposed simplified basis coefficients-frequency-BEM algorithm derives the mathematical expression between the basis coefficients and the channel frequency responses,so as to directly obtain the channel frequency response by taking advantage of the estimated basis coefficients.The two proposed algorithms avoid the high complexity of the time-frequency domain transform in traditional BEM,decreasing its computational complexity from??7?N ³?8?to??7?N?8?.According to the simulation results,when the velocity is 450km/h under the extended vehicular model,the performance degradation of the two proposed algorithm is just about 0.3dB as the BLER equals to 10^-4 compared with the conventional BEM algorithm,whereas error floor will appear for the interpolation-based method at BLER=0.1.2)Focusing on the fast time-varying channel estimation techniques for OFDM systems,we propose a novel simplified BEM-based channel estimation algorithm considering that the existing double one-dimensional?2×1D?interpolation algorithms have low estimation precision,and the iterative algorithms and the BEM algorithms have accurate estimation with high computational complexity in high-speed mobile scenarios.Utilizing the derived mathematical expression between the basis coefficients and the channel frequency responses,we can obtain the basis coefficients according to the estimated channel frequency responses,which differs with the second algorithm discussed in 1).Then,all the OFDM symbols'channel frequency response can be acquired directly according to the estimated basis coefficients.The proposed algorithm avoids large-matrix operations in the traditional BEM algorithm,thus the computational complexity is decreased to??7?N?8?.The simulation results show that the performance loss of the proposed algorithm is less than 0.1dB compared with the traditional BEM algorithm,while the performance enhancement is 1dB compared with the 2×1D interpolation algorithm.