With the development of electronic information society, big data of power grid has become the development trend of smart grid. The ever-increasing business of power grid has made the need for big data analytics more intense. The premise of big data applications is to ensure accurate time synchronization. For improving service quality and expanding business capabilities, various businesses and operations of the State Grid rely on accurate and efficient data processing and analysis. Traditional wireless time synchronization technology is not suitable for grid scenarios. Based on this, this paper starts with the multi-source grid network structure and analyzes the demand for wireless time synchronization of multi-source grid services. According to the requirements, this paper adopts a wireless physical layer-based multi-source grid wireless time synchronization mechanism and a Software Defined Radio (SDR) based network time synchronization system to meet the lowcost and high-precision requirements of multi-source grid for wireless time synchronization, which achieves multi-source grid wireless time synchronization with a worst-case accuracy of approximately 800 microseconds.