Abstract:  Compressed Sensing (CS) is a technique of sampling a sparse signal at a low rate and reconstructing it at the receiving end. The CS-based sparse channel estimation has smaller pilot overhead and better channel estimation performance. For the pilot design problem in CS-based OFDM sparse channel estimation, a new pilot location design method based on Tree-based Stochastic Search Algorithm (TSS) is proposed. Inspired by the structure of the tree，the proposed method performs a random search in the manner of branch tree to avoid falling into local optimum. Simulation results show that, compared with the traditional pilot design method, the pilot location obtained by TSS algorithm with lower complexity can obtain better channel estimation performance.