The Study on Channel Estimation Based on Superimposed Training Sequence for OFDM Systems
|Course||Signal and Information Processing|
|Keywords||Orthogonal Frequency Division Multiplexing (OFDM) Channel estimation Peak-to-Average Ratio(PAR) Superimposed training sequence|
Orthogonal frequency division multiplexing (OFDM) has been identified as the core technology of the fourth generation of mobile communication system due to its high spectral efficiency and robustness to frequency selective fading. As the constant demands of humans for digital communication, broadband, mobile, and personalized, OFDM technology is also being used widely in communication areas. OFDM systems usually adopts the multi-level modulation technology, and the receiver need the channel state information obtained by channel estimation for the coherent demodulation, so channel estimation is one of the key technologies in OFDM systems.In OFDM systems, channel estimation algorithms can be divided into three categories:pilot symbol assisted modulation channel estimation, blind channel estimation algorithm and semi-blind channel estimation algorithm. In recent years, domestic and foreign scholars proposed a new semi-blind channel estimation algorithm, superimposed training sequence channel estimation. In this paper, we do in-depth researches and analyses about the superimposed training sequence channel estimation. Paper contents are as follows:1. Studying the time-domain statistical average and frequency-domain statistical average based on the superimposed training. The simulation results show that: when using the same training sequence, the time-domain statistical average has higher efficiency of data power and better estimation performance; when using different training sequences, the time-domain statistical average obtains the best system performance with the sequence that has the constant modulus properties in time domain, and the frequency-domain statistical obtains the best system performance with the sequence that has the constant modulus properties in frequency domain; when the channel is the fast time-varying channel, the performance of the statistical average algorithm is poor. So it applies only to estimate the time constant or slowly varying channel. 2. To the APSB (added pilot semi-blind) algorithm of superimposed training sequence iterative decision channel estimation algorithms, an improved algorithm is proposed. The improved algorithm computes the channel’s correlation matrix with using the channel’s estimated value of the last iterative result. Simulation results show that:the improved algorithm do not rely on the channel characteristic, has certain robustness.3. A channel estimation algorithm based on selected superimposed training sequences is proposed. At the system transmitter, it makes some independent training sequences superimposed on the data sequence and selects one of the best PAR performances for transmission. At the receiver, it uses the corresponding training sequence for channel estimation. Simulation results show that: the algorithm achieves a good channel estimation performance and effectively reduces the system PAR without increasing the complexity of the system receiver.