Design and analysis of broadband amplify-and-forward cooperative systems: A fractionally-spaced sampling approach

In this paper, we propose novel fractionally-spaced frequency-domain equalizers for the amplify-and-forward cooperative systems. In the proposed equalization schemes, the sampling rate of the received signal remains at least as high as the Nyquist rate within the digital processing chain of the relay node(s), upon which a true fractionally-spaced equalization becomes feasible at the destination. Based on minimum-mean-square-error (MMSE) criterion, different equalization structures (linear and non-linear) are designed, and approximations for their bit error rate (BER) performance are presented. The BER performance of the proposed schemes are further lower bounded through a matched-filter bound (MFB) analysis which provides insight into system design such as optimum power allocation and relay selection strategy. Our results show that, under certain channel realizations and sampling phase errors (that may occur in the relay and destination terminals), the performance of the conventional symbol-spaced cooperative systems reduces to that of no relay scenario. However, the performance of cooperative systems with the proposed fractionally-spaced equalizers is independent of the samplers' phases, and as a result, full benefit of cooperation is retained.