The stochastic cloning Kalman filter (SCKF) accounts for the uncertainty of historical reference states related to relative observations. Compared with the extended kalman filter (EKF), SCKF demonstrates higher accuracy in processing relative observations and is widely applied in scenarios such as state estimation for mobile robots and unmanned aerial vehicle (UAV). However, as the number of relative observation sensors N (such as camera and LiDAR) increases, the dimensions of the state vector and covariance matrix in SCKF expand significantly. Since the computational load of operations like matrix inversion increases cubically with the matrix dimension, the computational complexity of the traditional algorithm rises drastically, which severely limits the application of the SCKF algorithm in embedded systems. To address this issue, this article proposes an improved Consider stochastic cloning Kalman filter (C-SCKF) algorithm, aiming to achieve efficient state dimensionality reduction and accurate estimation in scenarios with multiple relative observations. By introducing a state dimensionality reduction strategy and incorporating the error update mechanism of the consider Kalman filter (CKF), the proposed algorithm merely considers the uncertainty levels of the cloned states without directly estimating them. Consequently, the dimension of the state estimation is successfully reduced from 15(N+1), which originally scales with the number of relative observation sensors, to a fixed dimension of 15. Distinct from existing methods that only account for a single type of relative observation, this article derives a more generalized expression for SCKF capable of handling multiple asynchronous relative observations. Furthermore, the algorithmic complexities of EKF, SCKF, and C-SCKF are analyzed and compared under the condition of N types of relative observations. Numerical simulations and real-world experimental results demonstrate that, while maintaining a multi-source navigation accuracy comparable to that of the traditional SCKF algorithm, the computational speed of the C-SCKF algorithm is improved by approximately five times. The proposed algorithm effectively alleviates the computational burden caused by high-dimensional matrix operations and is more suitable for deployment on vehicle-mounted or airborne embedded devices with strict constraints computing power and power consumption.