Research On Channel Estimation Algorithm Based On Superimposed Training Sequence For OFDM

Orthogonal Frequency Division Multiplexing (OFDM) has advantages of high spectralefficiency and robustness against the multi-path fading. It has been widely concerned in thefield of wireless communications. And it was determined as a core technology of the fourthgeneration mobile communication systems. However, channel estimation is more difficultbecause of the frequency selective fading caused by multi-path effects and the inter carrierinterference（ICI）which is caused by channel time-varying characteristics. Therefore, anefficient channel estimation algorithm has become the key factor to the successful applicationof OFDM technology.Nowadays, there are three categories for channel estimation algorithms in OFDMsystems, just as pilot assisted channel estimation, blind channel estimation algorithm andsemi-blind channel estimation algorithm. Channel estimation based on superimposedsequence is belong to semi-blind channel estimation. In this paper, we do lots of research forthe algorithm as follows. A system model of superimposed pilot aided channel estimationbasic algorithm is introduced, and estimation algorithm based on various common trainingsequences are derived and simulation of the relationship between estimation performance andeach parameters is made, and the problem of PAPR and power are discussed.Single OFDM symbol channel estimation performance is poor, and this paper proposes anew advanced multiple-symbol estimation algorithm. The simulation results show that thealgorithm is effective to improve the estimate performance. From the mean square error ofchannel estimation based on superimposed training sequence we can see informationsequence is considered as noise, the serious impact on the accuracy of the estimation.Therefore, this paper introduces a new algorithm, whose main idea is to remove theinformation sequence on frequency bins before transmitter in order to eliminate informationinfluence, its performance is better than basic algorithm. But the introduction of newsequences of this method results in peak-to-average ratio increases, calculation of the meanmakes a certain delay to the system and valid information sequence’s power is relatively small,takes the receiver signal detection into trouble. To solve the above problem, this paperproposes a new algorithm. The algorithm principle is that the message sequence statisticalaverage is estimated effectively at the receiving end to avoid the introduction of new sequences. System simulation shows that the algorithm not only has the same estimation withthe original algorithm channel estimation, but also has greatly improved in bit error rate.