Tree-based optimized forward scheme for pilot placement in OFDM sparse channel estimation

The problem of channel estimation in Orthogonal Frequency Division Multiplexing (OFDM) systems Is here investigated. In particular, channels exhibiting a temporal response with sparse structure are considered. For fast fading channels, we consider the comb type estimation where we address the pilot allocation optimization by applying the recent Compressed Sensing (CS) theory. This optimization reduces to the choice of the CS decomposition basis, deduced from the Discrete Fourier Transform (DFT) matrix as a selection of rows associated with the pilot subcarriers. It has been demonstrated that the optimal CS basis, respecting the coherence measure minimization, corresponds to that based on Cyclic Difference Sets (CDS). In this contribution, for unavailable CDS, we propose a forward tree-based near-optimal pilot allocation scheme. This scheme, builds up iteratively the CS decomposition basis and minimizes the coherence measure in each iteration. It is shown that the proposed method attains a slightly enhanced performance compared to existing schemes yet with considerably reduced complexity which can be really beneficial in high mobility scenarios.