The Ofdm System, Frequency Offset Estimation Algorithm

With the development of communication techniques, communication systems with high quality, fast speed, and large bandwidth are more attractive for people. Orthogonal frequency division multiplexing (OFDM) is just viewed as one of key techniques for future wireless communications. It can both effectively combat multipath fading and improve bandwidth efficiency. However, OFDM systems are very sensitive to CFO (carrier frequency offset) who seriously degrades the performance of OFDM systems.In this paper, we focus mainly on the study of several kinds of CFO estimation algorithms, including their theory analysis, estimation range, and baseband simulation. A new algorithm is proposed for integer CFO estimation in OFDM systems with null subcarriers (NSCs). Our main works are as follows:1) For blind CFO estimation, we mainly focus on maximum likelihood(ML), multiple signal classification (MUSIC) and estimation of signal parameters via rotational invariance techniques (ESPRIT). From simulation and complexity analysis, we find: the bandwidth efficiency of blind estimation is high, but its complexity is usually higher, and its estimated range is limited; ML estimator has a good estimation accuracy and a lower complexity, but its range is limited to fraction CFO; MUSIC has a larger estimation range than ML, its performance is better than that of ML, but it is higher on complexity than ML; ESPRIT has the same estimation range as MUSIC and similar computational complexity, unluckily, in the case of high SNR, it is lower on accuracy than MUSIC.2) Considering that ML can not estimate integer CFO, we pay our main attention to the Sameer-Kumar (SK) method and X.MA method. Due to its poor performance for SK method and a high computational complexity for X.MA method, a novel MPS (minimum power sum) approach based on a distributed NSCs is proposed after the frequency selective property of wireless channels is considered. By simulation, it follows that the error detection probability of proposed MPS is much lower than SK method using the similar computational amount as SK. And the computational complexity is also much lower than X.MA method.