Slowly Time-Varying Channel Estimation Algorithm Based On OFDM And Its FPGA Implementation

Channel estimation is the key technology of the mobile communication. With theimprovement of data volume and transmission rate, requirements of channel estimationin communication system is becoming more and more demanding. This paper expoundsthe present situation of the study about channel estimation theory and discusses themain characteristics of mobile fading channel and classification. The channel’s commonmodel was introduced in the mathematical analysis and simulation experiment. Itfocuses on the instructions of Jakes model’s alternative model which is more suitable tothe frequency selective channel and complex-exponential basis expansion model(CE-BEM) which is commonly used in the time-varying channel simulation. In thispaper, the slowly time-varying channel estimation techniques based on OFDM systemsare studied and we put forward a new algorithm.Channel estimation using superimposed training (ST) adds pilots to theinformation data, which do not take up precious band resources and received extensiveattention. This paper fully introduces the research development and realization processof Channel estimation using superimposed training. Because channel estimation usingsuperimposed training has bad effect and data-dependent superimposed training channelestimation (DDST) is very complex, this paper give a special arrangement of pilotsadded in OFDM system and propose a new channel estimator using part-data-dependentsuperimposed training (PDDST), and apply it to the estimation of Slowly time-varyingchannel.Because pilots are impact sequence in the time domain, we cut a part of energyof the data which participates in channel estimation operation and near the pilots. Thismethod doesn’t have the shortcomings of DDST which is slow and computationalcomplexity. In the receiver, we propose the way to compensate of information data cutto avoid reducing data recovery performance since the information data amplitude is toosmall.Channel estimation using superimposed training and data-dependent superimposedtraining is computational complexity. So their applications are less in the actualcommunication system. In this paper, we give the implementation scheme of proposedalgorithm in FPGA platform and realize key modules of the algorithm.