Semi-blind channel estimation for block transmissions with non-zero padding

We focus on channel estimation algorithms for block transmissions that pad each information block with a fixed pilot block that is known at the receiver. For zero padding (ZP), such an algorithm (so-termed 'blind' algorithm) has been presented by Scaglione et al. (1998). In this paper, we present such an algorithm (so-termed 'semi-blind' algorithm) for nonzero padding (NZP). Similar to the existing blind algorithm for ZP, the proposed semi-blind algorithm for NZP operates on a group of received blocks and only requires a minimal redundancy of L symbols per transmitted block, where L represents a known upper bound on the channel order. However, while the existing blind algorithm for ZP can only identify the channel up to a scalar ambiguity the proposed semi-blind algorithm for NZP can fully identify the channel. Moreover, simulation results demonstrate that the proposed semi-blind algorithm for NZP outperforms the existing blind algorithm for ZP, even when the latter uses optimal (in practice infeasible) scalar ambiguity removal.