Constrained LS algorithm for channel vector estimation in 2-D RAKE receiver

In a conventional 2D RAKE receiver, the channel vector is usually estimated based on the LS criterion only using the pilot. When the power of the pilot is low or the variation of the channel is fast, the error of the channel estimation becomes large. We propose an improved algorithm in this paper. The steering vector of the desired signal is estimated firstly using a subspace decomposition method and a constrained condition is then configured employing this estimated steering vector. The channel vector is estimated based on the constrained LS criterion applying traffic signals. The pilot is only used to get the initial value of the estimation. The estimation accuracy of the channel vector can progressively increase through iteration. Simulation results demonstrate that the improved algorithm upgrades the estimation accuracy effectively.