In light of the equivalent joint sparse learning model

an effective method based on multitask Bayesian compressed sensing (MT-BCS) is presented for the design of pattern reconfigurable antenna arrays.The method can dynamically reconfigure arbitrary radiation patterns with the exact pattern details and as fewer number of antenna elements as possible.Firstly

the sparse learning model of multiple reference patterns is built based on MT-BCS theory and priori assumption about the priori probability of weight vectors.Then fast relevance vector machine (RVM) is exploited to estimate maximum posterior probability of hyper-parameter and further to obtain array optimizing positions and excitations.By varying excitations and optimized element positions

different patterns with desired and precise particulars can be achieved.Simulation results validate the efficiency of the proposed method for the design of maximally sparse reconfigurable antenna.