Abstract : State-of-the-art indoor broadband power line com- munications (PLC) systems use orthogonal frequency divi- sion multiplexing (OFDM) signals with constellations up to 10 bits/symbol, which makes channel estimation a key aspect. This paper focuses on the initial channel estimation used in the payload, which might be further adapted by means of a decision- directed strategy. This initial estimate has to be computed on a per-frame basis and can be accomplished from the preamble, the header symbol (assuming that it has been correctly decoded) or from a combination of both. This work proposes simple channel estimation techniques for this problem and derives their most appropriate parameters. They are compared with the linear minimum mean squared-error (LMMSE) estimator and others commonly used in wireless scenarios by considering both their performance and computational complexity. The factors limiting the performance of the estimators based on the preamble and the header symbols are analyzed and the implications of the differ- ences between PLC and wireless channels, such as the absence of fading, in the design of estimation techniques that require knowledge of the channel statistics are discussed. A performance analysis of the considered techniques is accomplished in a set of 171 measured indoor PLC channels. Obtained results indicate that estimators from the header perform better than those based on the preamble symbols and we provide a computationally simple estimator that gives nearly optimum performance by combining estimates from both the preamble and the header symbols.
Index terms : Broadband, channel estimation, OFDM, power line communications, preamble.