In this paper, we introduce a novel antenna positioning algorithm based on skeleton extraction. Indoor distribution of antennas is highly depending on the plane structure of the building, and thus skeleton extraction can be immediately applied to locating where antennas should be placed. We divided the plane structure of a building into several sub-areas, extracted the skeletons of all the sub-areas, and then positioned the antennas on the skeleton by circle covering method. We compared the proposed algorithm with the current method of utilizing manual site surveys, as well as with genetic algorithm based method. Simulation demonstrate that the average number of antennas used per 1 000m 2 decreased by 8.67% and 2.72% respectively but the average signal strength and the average coverage rate are still satisfy the requestment.