Positioning Using a Massive MIMO Software Defined Radio

This work focuses on positioning on a 2D plane using a massive MIMO software defined radio, yielding distance and angle of a movable mobile station relative to a stationary base station. Within the scope of this work, the software defined radio (built by National Instruments) was programmed to save channel estimates from real-world measurements of an 8 by 2 antenna MIMO OFDM transmission. The measurements were done outdoors and indoors, as to give results with and without significant multi path components and different environments. Hardware limitations, especially very small available bandwidth had to be taken in consideration.
Furthermore, a mathematical model of the signal processing in the hardware, the channel, received signals and estimation of the distance and angle was formulated.
This mathematical model was implemented and used with data from the measurements, yielding estimates for the two parameters.
Lastly, focus shifted on a statistical evaluation of the results, comparing them to theoretical lower limits of accuracy imposed by the Cramer-Rao lower bound.
Results showed that for two evaluated measurements the performance of distance estimation was sub-par, because the (unwired) synchronization between base station and mobile station was insufficient. In a third evaluated measurement, synchronization was done with an external cable, yielding results near the Cramer-Rao lower bound for distance estimates.
The angle estimation generally performed more reliable, but showed systematic errors for indoor localization, which could be attributed to reflections being dominant over line of sight components.