Updating of directions of arrival using training sequences in a GSM-based SDMA mobile radio system

In a space division multiple access (SDMA) mobile radio system, different mobile users use the same spectrum and are separated only by the directions of arrival (DOAs) of their signals at the base station. A critical task is therefore the estimation of the DOAs. When the users move around, the DOAs are time variable, which means that they must be tracked after their initial estimation. Using subspace methods (e.g. ESPRIT, MUSIC) is computationally expensive and the number of DOAs which can be tracked is limited by the number of antennas. In this paper an algorithm for updating the DOAs in a GSM-based SDMA mobile radio system is presented. It makes use of the training sequences used in the GSM system and updates the DOAs directly without using the subspaces. The proposed algorithm has low computational complexity and is not limited by the number of antennas. Monte Carlo simulations show that the tracking behavior of our algorithm is stable and that the error made is small, so that it represents a valuable complement to high-resolution DOA estimation algorithms, which can help reduce the hardware complexity and improve the real-time properties of the SDMA system.