To address the issues of the large errors in received signal strength indicator (RSSI)-based ranging methods and the limited adaptability of the iterative method based on the scaling by majorizing a complicated function (SAMCOF), which cannot adapt to the instability of distance measurements and degrading localization precision, this paper proposes an adaptive weighting algorithm based on the extended Kalman filter (EKF) and multidimensional scaling (MDS). The algorithm first fuses the distance measurements obtained from RSSI and acceleration information using EKF to obtain optimized distance states. Then, the weights for different communication node pairs are dynamically adjusted based on the confidence of the distance states in the covariance matrix, and an optimized distance matrix is constructed for MDS-MAP positioning to obtain the initial positions. Finally, the SMACOF-based iterative optimization method is employed to refine the initial positions, reduce the negative impact of incomplete link observations and enhance positioning accuracy. Simulation experiments show that the proposed localization algorithm outperforms MDS-MAP, vMDS, and wMDS in various network distributions, communication radii, node numbers, and noise levels, improving positioning precision and robustness in dynamic networks. Additionally, the semi-physical experiment results of the positioning system based on ZigBee CC2530 validate the effectiveness of the algorithm’s effectiveness in both indoor and outdoor scenarios, overcoming the limitations of traditional methods in complex environments.