Channel estimation

Abstract

Channel estimation is a mandatory task at the receiver in coherent Orthogonal Frequency Division Multiplexing (OFDM) systems. It must be carried out with the use of some channel information, and this channel knowledge can be obtained with either data-aided methods, where some pilots or preambles are sent with the information, or decision-directed schemes that only make use of the decisions on information symbols by blind schemes. In general, data-aided methods are preferred, that are based on inserting known symbols (pilots) in the time-frequency grid with a certain pattern. First, the channel is estimated at pilot positions and, at a second step, the estimation algorithms are applied to interpolate the channel frequency response at data positions, based on the channel knowledge at pilot positions. Efficient Discrete Fourier Transform based methods can be employed for the interpolation in the frequency dimension. Also, Support Vector Machines have been proposed for robust channel estimation in the presence of impulse noise. In multi-antenna configurations, the channel estimation requires a careful design to ensure orthogonality of the preambles among the different transmit antennas. This guarantees that different sub-channels of the Multiple-Input Multiple-Output OFDM system can be resolved at the receiver. Recently, there is a great interest in the reduction or suppression of the cyclic prefix to increase capacity. This leads to efficient systems that however must deal with inter-symbol and inter- carrier interferences. In that case, channel estimation strategies might rely on symmetric sequences. The multi user scenario found in the uplink Orthogonal Frequency Division Multiple Access systems is also addressed. The particular details of how channel estimation is implemented in certain standards such as WiFi and Wi max are summarized, with special emphasis on the specifications for the preambles. Finally, some conclusions and future lines of research are envisaged. © 2012 Nova Science Publishers, Inc. All Rights Reserved.