Aperiodic array pattern synthesis

The steepest descent algorithm (SDA) is very attractive for finding a local optimum when a problem can be formulated in a mathematical equation. The author used SDA in optimizing the sidedobe characteristics of an aperiodic arrays consisting of 16 sub-arrays (N. Toyama, 2005). The optimum solution obtained by SDA is fully dependant on the initial value used as an input of SDA program. Therefore, creating initial values are very important to find the best solution for a given problem. The author have proposed two methods, namely, a form of linear equation with several parameters which is used in the reference 1, and a binary code representation method where binary codes are selected appropriately (N. Toyama, 2006) . In this paper the author proposes a new method for giving initial positions, where random numbers created by a computer are used to give the 16 sub-arrays´ initial positions. The author calculated the side-lobe patterns again and again by using SDA with initial positions given by the new method. The result so far obtained shows about 2 dB improvement in the side-lobe patterns compared with the other two method given above.