Conclusions and future research directions

In this book, we have discussed channel estimation for various situations in PLNC, including frequency flat fading environment, frequency selective fading environment, and time-selective fading environment. In each environment, we demonstrated how the channel estimation is different from the conventional point-to-point transmission as well as from the uni-directional relay network. The key idea is that the individual channel knowledge should be obtained at three nodes, i.e., the terminals and the relay, within one round of the data exchange. One may, of course, apply more complicated training process, say, training each channel link separately and share the information through feedback channels but such processing is not compatible with the overall structure of the data frame. Moreover, we developed channel estimation algorithms that fit the speciality of different environments, for example in frequency selective fading environment it is possible to remove the redundant estimates so that the overall training length could be reduced, while in time selective fading environment the individual BEM coefficient is estimated instead of the channel parameters.