Employing evolutionary optimization for designing practical beamsteering algorithms for the uplink of UMTS with transmit power control bits as feedback

Abstract

So called open loop transmit diversity algorithms for the uplink of UMTS/HSPA use the power control commands of the fast power control mechanism as a feedback information to iteratively determine the antenna weights, thus making them compatible with existing networks and procedures. Although these algorithms are a topic of recent studies within the 3rd Generation Partnership Project (3GPP), the design methodology of developing such algorithms has not been covered in literature so far. Such a systematic design method is the topic of this paper. First, a general model is developed where the beamsteering algorithm is represented as a finite-state machine that takes the transmit power control commands as input and delivers the antenna phase as output. Then, multiobjective evolutionary optimization of these state machines is utilized in order to find optimal beamsteering algorithms. The objectives during this optimization comprise not only the reduction of transmit power at the User Equipment but also the minimization of the irritation of the UMTS fast power control loop by the iterative beamsteering algorithm. As the result of the optimization, a beamsteering algorithm is found that reduces the transmit power at the User Equipment by 3.31 dB compared to the baseline case of employing only one transmit antenna. The influence of the number of discrete phase quantization steps on the performance of the algorithm is also studied and it is found that quantizing the relative antenna phase in 22.5 degree steps is sufficient.