Magnetometers are widely used for attitude determination of low Earth orbit (LEO) information satellites. In order to estimate the satellite attitude accurately, bias of magnetometer must be estimated. In this study a Linear Kalman Filter (LKF) based algorithm for the estimation of magnetometer biases is proposed. Proposed algorithms are simulated through attitude dynamics of a small satellite. We shortly present also discrete algorithms for in-flight calibration and alignment of a low cost strap-down inertial navigation system with correction by signals from the Sun and magnetic sensors.