Channel estimation for indoor power-line communication systems

An indoor power-line channel features frequency-selective fading, which results from the tree-like structure of the power-line network and the various loads connected to its termination points. Power-line channel estimation yields the channel state information that is usually indispensable for evaluating and compensating the degradation suffered by the signal propagating over power lines. There are two different categories of channel estimation strategies. The first one is to utilize the physical characteristics of the power- line network, based on which, the channel is identified using a channel modeling technique; the second one is to regard the channel as a "black box" and estimate its status by exploiting the change in the signal's characteristics after its propagation in the channel.;This research focuses on the development of two original channel estimation approaches, each belonging to one of the categories mentioned above. The first approach is based on a power-line-channel time-domain modeling strategy. With the knowledge about the power-line network's topology, cable characteristics and load conditions, this algorithm utilizes simple but effective recursive matrix operations to accurately compute the channel's frequency response at a given frequency band. Using this algorithm, the frequency response of any point-to-point channel in a power-line network can be calculated in an efficient manner. The effectiveness of this method is demonstrated by comparing the estimated frequency responses of two example power-line networks with the corresponding measurement results. The second channel estimation approach developed in this research is to identify the channel by exploiting the orthogonal frequency division multiplexing (OFDM) signal's second-order statistics (SOS) through the use of a subspace decomposition procedure. Different from the previous work on channel estimation in OFDM systems, the proposed algorithm is specially derived for OFDM with window shaping, which is commonly exercised in practice for better controlled bandwidth. A modification in the window shape and its influence on the estimator's performance are also analyzed. The scalar indetermination resulting from the subspace decomposition procedure is resolved by taking advantage of the properties of subchannel constellations and power-line channels. Therefore, totally blind (i.e., pilotless) channel estimation is achieved. Simulation results have demonstrated the effectiveness of the proposed algorithm.