Research Of Noise Variance Estimation Algorithm Of LTE System And The Realization On Multi-Core DSP

With the increase of demand on the multimedia services,3GPP has proposed its long term evolution (LTE) system. The fundamental targets for LTE concern increased data rates; reduced latency, improved system capacity and coverage as well as reduced operator costs. The most significant improvement of LTE system is that it employs many novel air-interface techniques. Accordingly exact noise variance estimation technology not only is the basis for MMSE equalization and soft demodulation but also is the key to the high transmission performance and ensures the superiority of the system. So this thesis is engaged in the research on the noise variance estimation technology for LTE.This work mainly contains the following three aspects. First, six classic algorithms of noise variance estimation are researched. Including the M2M4 estimation algorithm, Boumard's estimation algorithm, maximum likelihood algorithm, minimum mean square error algorithm, frequency domain noise variance estimation algorithm and time domain noise variance estimation algorithm. As well as analysis and comparison of various algorithms, combined with LTE system requirements, taking into account factors such as complexity, hardware resources needed and latency. Select the time domain noise variance estimation algorithm.Second, this thesis gives the performance curves of the time domain noise variance estimation algorithm. The simulation results show that the performances of time domain noise variance estimation algorithm are close to the ideal noise variance estimation.At last, from the perspective of implementation, the process of the noise variance estimation module in LTE uplink is introduced, as well as the DSP platform for the implementation. Moreover, the hardware implementation method of IDFT sub-module is introduced in detail.