978-1-4799-2764-7/13/$31.00 ©2013 IEEE 1221
2013 6rd International Congress on Image and Signal Processing (CISP 2013)
A Novel Unitary Beamspace Recursive Least
Squares Algorithm for Fast Adaptive Beamforming
Wei Shao, Ting-hui Yin, Fan-qiu Meng, Zu-ping Qian
College of Communications Engineering
PLA University of Science and Technology
Nanjing, China
Abstract—A novel unitary beamspace recursive least squares
(UB-RLS) algorithm for fast adaptive beamforming is proposed
in this paper. Firstly, a kind of random unitary transformation
matrix is designed. Secondly, with these matrices, the feed signals
are transformed into a low-dimensional space named the unitary
beamspace (UB) and the computation complexity is accordingly
decreased. Finally, the RLS algorithm is applied for the
beamspace to realize adaptive beamforming and the whole
processing is called the UB-RLS algorithm. As a result, a good
signal of interest (SOI) extraction and interferences nulling
performance is obtained. Simulation experiments show the merits
of the proposed algorithm against the other two, the UB-LMS
algorithm and the UB-MVDR algorithm.
Keywords-unitary beamspace recursive least squares (UB-RLS)
algorithm; adaptive beamforming
I. INTRODUCTION
With good abilities of anti-jamming and system capability
enhancement, the adaptive beamforming technology is widely
applied in many fields as wireless communication[1][2],
satellite navigation[3] , radar[4] and etc. However, the burden
of computation complexity as well as cost and power
consumption gets much heavier when the number of array
elements is big. Therefore, the beamspace processing[5] and
fast adaptive algorithm play an important role on solving the
above problem.
The computation speed for beamforming is related to
several factors such as calculation complexity and convergence
speed. Recently, an unitary beamspace (UB) processing[6][7]
is presented to transform the feed signals from high-
dimensional element-space into low-dimensional beamspace
through a series of random unitary matrices, which effectively
decreases the computation complexity of the beamforming. At
the same time, different adaptive beamforming algorithm has
respective convergence character. With a fast convergence and
good anti-jamming performance, the RLS algorithm is
attractive for adaptive beamformer design. Therefore, with
advantages of low complexity and fast convergence from the
UB processing and the RLS algorithm, a novel UB-RLS
algorithm is proposed in this paper. Also, the performance
comparison is made among the proposed algorithm, the UB-
LMS algorithm and the UB-MVDR algorithm.
This paper is organized as follows. In section II, the signal
model is introduced. In section III, a novel UB-RLS algorithm
is proposed. In section IV, we present some simulation results
illustrating the performance of the proposed algorithm. Finally,
in section V, we give some conclusions.
II. S
IGNAL MODEL
Figure 1. Liner uniform array
As shown in Figure 1, consider an uniform linear array with
N array elements separated by a spacing of