The Algorithm Of UV-LMS And TRLS For Beam-forming Of Smart Antenna

The core of smart antenna research is the arithmetic of beaming. Yet some of the arithmetic have the problem of needing lots of data, high computational complexity and slow speed of convergence. How to improve the arithmetic in order to improving performance is one of the most important research point. This article researches on the Least Mean Squares (LMS) arithmetic and recursive least-squares (RLS) arithmetic with the purpose of reducing total convergence time. The work can be divided into five aspects as follows:1. This paper analyzes the influence of vector's correlation to the LMS algorithm. This paper introduces a uncorrelated LMS (U-LMS) algorithm which speeds up the convergence without adding malajustment.2. Based on the research of contrary between fixed-step and maladjustment of LMS algorithm and U-LMS, this paper introduces a uncorrelated variable-step LMS (UV-LMS) algorithm which makes the convergence speed faster at the beginning and the maladjustment is smaller at the end.3. This paper introduces a tree-model RLS (TRLS) to solve the problem of RLS's high computational complexity by splitting inputting data and treat it ina layered way which reduce the computational complexity from O(N ) to 0(N).4. The influence to the algorithm caused by the correlation of input data is discussed, which shows that low correlated inputting data lead to small maladjustment. In order to get small maladjustment in high correlated inputting data situation, we could pretreat the data with DCT (discrete cosine transform).5. The two improved algorithms' efficiency has been approved by simulation. UV-LMS has faster convergence speed and lower maladjustment than LMS and TRLS has shorter convergence time than RLS which makes them suitable using in dynamic fast changing environment and improving the whole system's efficiency.