Robust path detection for the LTE downlink based on compressed sensing

The authors will present an overview and simulation results of channel estimation exploiting sparsity for the concrete parameters of the long term evolution (LTE) system. Recent progress in compressed sensing has shown that O(s log(p)4) randomly placed pilots are sufficient to recover a s-sparse impulse response of length p by solving a convex relaxation of the combinatorial sparsity problem. The pilot configuration in LTE is already standardized and a support for random pilot placement is not included. But the authors show that also in this fixed and deterministic setting substantial gains have to be expected from non-conventional channel estimation that make explicit use of the sparse nature of mobile communication channels. They have evaluated methods of channel estimation exploiting the sparsity of mobile communication channels. They have observed that current theoretical bounds are still too weak and not applicable to the usual LTE setup. However, under moderate conditions as present in typical LTE environments, the authors have obtained substantial gains with channel estimation based on sparsity over conventional linear minimum mean squared error (LMMSE) methods. Furthermore, the proposed channel estimator keeps almost stable when reducing the pilot overhead. This might be an important detail for pilot assignment and sequence planning in the multicell context.