In indoor positioning scenarios, the deployment of anchor nodes has a very important impact on positioning performance. We consider an indoor environment with obstacles, where it lacks environmental details or lacks equipment to simulate indoor signal propagation. We comprehensively consider the locating rate, positioning accuracy, and whether to deploy redundantly to design a new evaluation function. Then we use an improved gravitational search algorithm to solve the problem, which can obtain a better solution to the indoor positioning anchor node deployment problem within an acceptable time cost. The simulation experiment consists of two parts. The first part compares the algorithm with three reference algorithms. It has better performance in terms of positioning rate, geometric dilution of precision, positioning error and so on. The second part compares the evaluation function considering some factors. The analysis of multiple index items proves that the proposed evaluation function can obtain a better anchor node deployment scheme, which can effectively improve the anchor node topology and positioning performance.