Orthogonal training signal relaying for channel estimation in dual-hop AF relay networks

In this paper, an orthogonal training signal relaying technique for channel estimation in a dual-hop amplify-and-forward (AF) relay network with multiple relays is proposed. In an existing technique, the cross interference occurs between the precoded training signals of the relays since their precoded noise vector terms are not orthogonal. This leads to an increased channel estimation mean squared error (MSE) at the destination. To tackle the cross interference problem without changing the channel estimator at the destination, we propose to use an additional step at each relay prior to training signal precoding. That is, converting the received training signal vector to a scalar channel estimate term via estimating the channel of the first hop at each relay. In this case, the received noise vector at each relay is converted to an scalar channel estimation error term. When multiplying the mutually orthogonal precoding vectors with the scalar values (i.e. the estimated channels at the relays), the resultant precoded training signals of the relays remain mutually orthogonal. Hence the cross interference is eliminated. The proposed technique can provide a significantly lower channel estimation MSE at the destination as the number of relays M increases. Results show that the proposed technique yields 0.5 dB, 1.5 dB and 2 dB of BER performance gains over the existing method at a bit error rate (BER) of 0.001 when M are 2, 4 and 8 respectively.