Abstract
Hardware complexity reduction is critically important for Rake receivers in ultra-wide band systems, where the number of channel multi-path components (MPCs) can be very large. Though several designs (S-Rake, P-Rake) have been proposed, the impact of channel estimation overhead has not been accounted for in prior literature. In this paper, we propose a new Judiciously trained Selective Rake receiver that performs channel estimation for only a subset of the many channel multi-path components. The intuition behind the design is that only a few channel MPCs have enough strength to make them sufficiently likely to contribute significantly to the received signal power. By restricting the channel estimation to only these MPCs, the estimation overhead can be reduced significantly without degrading capacity. In this paper, we first show that. if the MPCs have independent Rayleigh amplitude fading, it is data rate optimal to perform channel estimation for only a subset of the MPCs with the highest second-order moments. We also present a closed form expression for the achievable data-rate of a JS-Rake receiver in Rayleigh fading and find the data-rate maximizing size of this subset. Simulations in several practically relevant channel settings show that JS-Rake boosts data rate by 30 - 40% and 50%, in comparison to S-Rake and P-Rake, respectively.