The authors investigate the performance of amplify-and-forward relaying with two different pilotsymbol-assisted channel estimation methods. In the first estimation method, the cascaded channel consisting of source-to-relay and relay-to-destination links is estimated at the destination terminal. No channel estimator is required at the relay terminal. In the second estimation method, the estimation of cascaded channel is disintegrated into separate estimations of source-to-relay and relay-to-destination links which are carried out at the relay and the destination terminals, respectively. The latter method involves feed-forwarding a quantised version of the source-to-relay channel estimate to the destination terminal. Through the derivation of mean squared error (MSE) diversity gain analysis and Monte-Carlo simulations, the authors investigate the performance of amplify-and-forward relaying with these channel estimation methods. Our results demonstrate that full diversity can be achieved in the presence of channel estimation. The authors further show that cascaded channel estimator outperforms its competitor with a small number of quantisation bits. As the number of employed quantisation bits increase, disintegrated channel estimator approaches to its competitor eventually outperforming it.

Abstract-In this paper, we investigate the error rate performance of amplify-and-forward (AF) relaying with imperfect channel estimation. We consider a single-relay scenario with orthogonal and non-orthogonal AF (OAF and NAF) cooperative protocols. Two pilot-symbol-assisted receiver architectures are studied: In the mismatched-coherent receiver, the complex fading channel coefficients (i.e., both phase and amplitude) are estimated based on a linear minimum-mean-squared-error estimation approach and fed to a coherent sub-optimal maximum likelihood decoder as if the channels were perfectly known. In the partiallycoherent receiver, channel amplitude is ignored and phase is estimated by a phase locked loop. For both receiver types, we analyze the achievable diversity orders for cooperative protocols under consideration and quantify the impact of channel estimation through the derivation of pairwise error probability. Our performance analysis reveals that a second order diversity order is obtained for the considered single-relay scenario indicating that full diversity is extracted. Our simulation results demonstrate that the performance degradation due to channel estimation with respect to the genie bound (i.e., perfect channel state information) is as small as 1.1dB based on the employed detector. Performance results further show that partially-coherent receiver presents a similar performance to mismatched-receiver for sufficiently large loop SNRs although channel amplitude is completely ignored.

No abstract

Abstract-Cooperative communication techniques promise the advantages of multi-input multi-output (MIMO) communications for wireless scenarios with single-antenna terminals. A main assumption in majority of the research work on cooperative communications is the availability of channel state information at the receiver. In practice, knowledge of the channel is obtained by sending known training (pilot) symbols to the receiver. In this paper, we study the effect of training on the performance of an amplify-and-forward cooperative relaying system with pilot-assisted channel estimator over quasi-static Rayleigh fading channels. We consider average received signal-to-noise ratio at the destination node as the objective function and formulate optimization problems for a single-relay scenario under total network power (TNP) and individual node power (INP) constraints. We aim to answer the following fundamental questions: Q1) How should overall transmit power be shared between training and data transmission periods?; Q2) How should training power be allocated to broadcasting and relaying phases?; Q3) How should data power be allocated to broadcasting and relaying phases? Our simulation results demonstrate that optimized schemes significantly outperform the original schemes with equal power allocation. Depending on the relay location, performance gains up to 5.5 dB and 2.8 dB are observed, respectively, under TNP and INP constraints.

No abstract