To measure the physiological parameter accurately, a plenty of sensors are placed in, on or around the human body, which form the Wireless Body Area Network(WBAN). But for energy saving, some sensors can be chosen to achieve the same goal. So an iterative convex optimization algorithm with constrained beam-pattern synthesis of sensor selection is proposed to solve this problem. In the optimization procedure, the objective is to find a tradeoff between the array weight which denotes the sparsity of the array and the Peak Side lobe Level(PSL) which denotes the radiation performance of the sensor array. The method achieves an excellent tradeoff between the array sparsity and energy efficiency. The cuboid sensor array in three-dimensional space is investigated and implemented on MATLAB. Numerical simulation results demonstrate that the algorithm is efficient to synthesize sparse cuboid sensor array, and at the same time minimize the PSL.