An efficient grid-based RF fingerprint positioning algorithm for user location estimation in heterogeneous small cell networks

This paper proposes a novel technique to enhance the performance of grid-based Radio Frequency (RF) fingerprint position estimation framework. First enhancement is an introduction of two overlapping grids of training signatures. As the second enhancement, the location of the testing signature is estimated to be a weighted geometric center of a set of nearest grid units whereas in a traditional grid-based RF fingerprinting only the center point of the nearest grid unit is used for determining the user location. By using the weighting-based location estimation, the accuracy of the location estimation can be improved. The performance evaluation of the enhanced RF fingerprinting algorithm was conducted by analyzing the positioning accuracy of the RF fingerprint signatures obtained from a dynamic system simulation in a heterogeneous LTE small cell environment. The performance evaluation indicates that if the interpolation is based on two nearest grid units, then a maximum of 18.8% improvement in positioning accuracy can be achieved over the conventional approach.