Feedforwarding Channel State Information in Relay Networks
Cenk M. Yetiş, Ahmet Hamdi Kayran · arXiv (Cornell University) · 2008
Channel estimation via training is a common method for providing imperfect channel state information (CSI) in a network. In a relay network, imperfect CSI can be efficiently provided at the relays by sending training signals from the source and the destination nodes to the relays. Providing imperfect CSI at the destination is more challenging since each relay is required to send training signals to the destination. We introduce the feedforwarding method, where each relay simply forwards the imperfect CSI to the destination in such a way that we benefit from the relay network structure to carry the information for us. This method significantly lowers the network resources spent on providing the imperfect CSI at the destination. In addition, we prove that the feedforwarding method provides better average effective signal to noise ratio (eSNR) compared to the training methods currently known in the literature. The eSNR is an important parameter in the channel capacity expressions. Therefore, the parameters of a network with training could be optimized in order to maximize the eSNR. In our setup, source, relay and destination nodes have M, N and M antennas, respectively. We use matched filtering at the relays and single antenna decoding at the destination, where each source antenna communicates to a separate destination antenna via K relays. We achieve our results at large K. Index Terms Relay networks, imperfect channel state information (CSI), channel estimation, effective SNR, capacity, training.