Low-Dimensional SDP Formulation for Large Antenna Array Synthesis

Bounding the sidelobe and mainlobe levels of an array with complex weights is attractive in that it allows direct control of the radiation pattern. In this paper we propose an efficient beamforming technique for synthesizing the patterns of large arrays with bound constraints on the sidelobe and mainlobe levels. It is shown that the pattern synthesis problem can be posed as a convex semi-infinite program (SIP) which is then turned into a semi-definite program (SDP) via a novel linear matrix inequality (LMI) characterization of semi-infinite trigonometric polynomial constraints. In contrast to existing SDP formulations which require a large number of additional variables, our SDP representation only uses a minimal number of additional variables. This subsequently enables the design of patterns for arrays with several hundred elements to be efficiently achieved on a standard personal computer using existing SDP solvers.