A method for measuring a geographical position of a device based on a network of satellites in a predictive system with a filter of variable gain, said gain being represented by a vector of variable gain coefficients, said method, implemented by the device, comprising at each iteration following a time t the steps of:
- receiving signals transmitted by a plurality of satellites in the satellite network,
- determining the position and/or the speed of the device at a time from the signals received, referred to as “observation at time”,
- calculating the best prediction of the state of the system at time on the basis of a predetermined estimation of the state of the system at time and a model representing the system between time and time,
- calculating the prediction of the observation at time as the product of a predetermined observation matrix and the best prediction of the state of the system at time calculated,
- calculating the innovation of the filter at time as the difference between the observation at time and the prediction of the observation at time,
- determining the vector of gain coefficients at time by minimizing the square of the norm of the innovation of the filter at time calculated in the previous step,
- calculating the gain at time by making a correction by stochastic approximation using time averaging of the gain coefficients of the vector of gain coefficients determined,
- calculating the estimation of the state of the system at time as being the sum of the best prediction of the state of the system at time calculated and the product of the gain at time by the innovation of the filter at time calculated,
- determining the position of the device from the estimation of the state of the system at time calculated.