Hide/Show Apps

Optimization techniques for nonconvex problems and optimum discrete transmit beamformer design /

Download
2014
Demir, Özlem Tuğfe
In this thesis, transmit beamformer design is investigated for single group multicast scenario. The problem is considered for both discrete and continuous case. The discrete problem is converted to a linear form in which there are both discrete and continuous variables. The resulting mixed integer linear programming problem is optimally solved with much lower computational complexity than brute force search. For practical reasons, robust version of the problem is also elaborated and solved with mixed integer convex programming. Several experiments are carried out in order to show performance gain and computational complexity of the proposed techniques. An important variation of discrete beamforming problem for spectrum sharing based cognitive radio is also considered. Antenna and secondary user selection which are critical in cognitive radio scenario are included into this beamforming problem. An equivalent problem to this joint problem is obtained and solved optimally using mixed integer linear programming. It is shown that antenna selection provides the system with power gain and more user service capability. Finally, a near-optimal continuous broadcast beamforming algorithm based on alternating maximization is developed and its performance is shown to be better than the existing approaches in simulation results.