Partially blind identification of FIR channels for QAM signals

For reliable communication equalization is necessary at the receiver. Conventional adaptive schemes rely on a known training signal to probe the unknown channel. In contrast, blind identification/equalization techniques do not require a training sequence, using instead the known statistical properties of the transmitted signal and observed statistics of the received data to estimate the channel response and/or the necessary equalizer. Our goal is to try to improve the conventional adaptive identifier by combining these approaches. Since, the resulting adaptive identifier uses both, a short training sequence and the properties of the transmitted signal to estimate the channel response, it is called a semiblind or a partially blind technique. This can also be looked at from the adaptive system identification point of view where both the direct (training signal) and indirect (data statistics) knowledge about the system are used for better system identification.