Considering the energy harvesting properties of solar-driven wireless sensor network node, an optimal transmission power scheduling policy was proposed aiming at maximizing data bits delivered by a deadline. Assuming data bits have arrived, the proposed forward-backward searching (FBS) transmission power policy established an optimal transmission power series. According to the average harvested energy estimate during the harvesting decreasing period, the method can provide a global optimal transmitting series. Simulation was made based on the solar radiation data of Los Angeles(N33°,W118°), America in 1990. The results show that FBS performs better than the energy used up(EUP) policy. The longer sensor node harvestes energy , the better FBS performs.