History inertial measurement unit (IMU) dates back to 1930. At that time, the technology because of its serious limitations on size, cost and power consumption, and the public was not applicable to small devices. Recent developments inertial measurement unit of MEMS sensors that are lighter, smaller and cheaper than other types of systems are inertial. These have made the inertial sensors in a variety of small devices such as mobile phones placed today. In this article, the first to introduce a variety of inertial systems and focus more on MEMS systems has been based on the mathematical model used where errors bias, scale factor and non-orthogonal axes is located. In summary, each of these errors can be expressed as follows. bias is the measurement error, regardless of the forces or the input rate of the sensor. The scale factor expresses the scale of the relationship between the output of a sensor with the force or the rate applied to the sensor. In other words, this error shows the difference between the ideal observation and the output of the sensor, and linear gradients are generated in observations in the inertial navigation systems. non-orthogonal axes between the axes is a defect in the construction and a lack of alignment of the three axes of measurement in the accelerometer or gyroscope. The following guidelines are provided to calibrate the static inertial systems. The methods by six static position (SPS), multi-position (MP) and recursive least squares (RLS) is an IMU used to determine the parameters of error. In this research, OARS software is used, a data aggregation system developed on the Android operating system. With 100Hz, this software provides accelerometer, gyroscope, magnetometer and GPS observations at 1 Hz. Observations of these sensors are stored in the csv format software in the phone's memory. In this study only gyroscope and accelerometer sensors are used. The results showed that the proposed methods First, the mathematical model with 12 parameters high dependency between the parameters compared to model 9 parameters. Secondly, if the raw observations in a particular situation, and the values averaged better results than entering into the equation raw observations or observations been removed noises by discrete wavelet transform. The unit of measurement error values in the multi-position on the three modes obtained by 0.082345, 0.083140 and 0.082952  is. Thirdly, the results of the error measurement unit in three ways proposed by 0.146245, 0.161520 and 0.082345  and because of the need to collect data in particular (exact alignment inertial system) to determine the calibration parameters show that the multi-position method is better than the other two methods.