Abstract: Indoor positioning is an important research field in current intelligent construction. The centroid algorithm is characterized by its simplicity and low computational cost for indoor positioning. However, it suffers from poor positioning accuracy and stability. To address this issue, an improvement is proposed by utilizing the difference in signal strength (RSSI) in the centroid algorithm. The improved centroid positioning algorithm based on RSSI difference requires deploying the reference nodes involved in positioning as a localization area in the form of an isosceles right triangle. The idea of narrowing down the localization area is employed by dividing the area evenly into four sub-areas along the perpendicular bisectors of the triangle's vertices. During the positioning process, the three reference nodes with the strongest signal strengths are selected, and it is checked if they form an isosceles right triangle. If the condition is met, the sub-area where the target node is located is determined by calculating the signal strength differences between the vertices of the reference node triangle. Finally, the centroid coordinates of the focused sub-area are computed as the final positioning result. Simulation results demonstrate that the improved algorithm outperforms the traditional centroid positioning algorithm and the three-side centroid algorithm in terms of higher positioning accuracy and better stability. Moreover, the improved algorithm avoids the need for using path loss functions, reducing the deployment workload of reference nodes in the early stages. Therefore, the improved algorithm is a relatively simple, practical, and highly accurate and stable indoor positioning algorithm suitable for various indoor wireless positioning systems.