On the Efficient Beam-Forming Training for 60GHz Wireless Personal Area Networks

In this article, we suggest an efficient beam switching technique for the emerging 60GHz wireless personal area networks. Given the pre-specified beam codebooks, the beam switching process, aiming to identify the best beam-pair for data transmissions, is formulated as a global optimization problem in a two-dimension plane that is formed by the potential beam pattern index. As the analytical gradient information of the objective reward function is practically unavailable, Rosenbrock numerical algorithm is properly adopted to implement beam searching, by implicitly approaching and exploiting the gradient descent direction through the numerical pattern-search mechanic. In order to enhance search performance, furthermore, a novel initialization process is presented to provide the feasible initial solution for Rosenbrock search. Inspired by the appealing conception of small-region dividing and conquering, this pre-search algorithm can efficiently reduce the search scope and hence improve the success probability. The developed beam switching technique, i.e. an initialization process followed by Rosenbrock search, exhibits a much lower complexity than the current state-of-the-art strategies. It is demonstrated from both theoretical analysis and numerical experiments that, compared with the existing popular methods, the required protocol overhead of the new beam-training procedure can be significantly reduced, accompanying the power consumption of 60GHz devices.