1796 IEEE COMMUNICATIONS LETTERS, VOL. 20, NO. 9, SEPTEMBER 2016
A Spherical-Wavefront-Based Scatterer Localization Algorithm
Using Large-Scale Antenna Arrays
Jiajing Chen, Stephen Wang, and Xuefeng Yin
Abstract—In this letter, a spherical-wavefront-based
parametric signal model is proposed by considering the
architecture using a large-scale antenna array. A space-
alternating generalized expectation-maximization (SAGE)-based
localization algorithm is derived and adopted to localize the first-
and last-hop scatterers along the propagation paths between
the transmitter and the receiver. Simulations are conducted
to evaluate the performance of the localization algorithm.
An outdoor measurement campaign with a 1600-element
virtual planar array is carried out to verify the localization
performance in reality. Those results show the advantages
of using a spherical-wavefront-based signal model and the
SAGE-based algorithm in the scatterer localization.
Index Terms— Spherical wavefront, SAGE algorithm, scatterer
localization, and large-scale antenna array.
I. INTRODUCTION
L
ARGE-SCALE antenna arrays challenge the validity of
the planar wavefront based signal model in the fifth gen-
eration (5G) wireless communications. In particular, a planar
wavefront is only observed when the sources are far away from
a small-scale antenna array. However, in reality many scatter-
ers are located close to the antenna array, especially in indoor
scenarios where the distance between a scatterer and a trans-
mitter (Tx) a receiver (Rx) array are smaller than the Rayleigh
distance. The indoor propagation channel consists of multipath
components (MPCs) and in many cases, with no clear line-of-
sight (LoS) components, which makes accurate indoor local-
ization challenging and increases the necessity of applying
novel localization approaches in algorithm design. The spheri-
cal wavefront assumption characterizing the sources with finite
length from the array is more accurate than the planar wave-
front when large-scale arrays are implemented. In addition,
the estimation of distance from the nearest scatterers to the
array is only feasible by adapting the spherical wavefront
assumption. Combined with the estimated angles of arrivals
and angles of departure, the spherical wavefront assumption
has its unique advantages in the scatterer localization. In the
case where the beamforming and beam-tracking techniques are
Manuscript received April 27, 2016; revised June 17, 2016; accepted
June 23, 2016. Date of publication June 28, 2016; date of current version
September 8, 2016. This work is jointly supported by the Natural Science
Foundation of China (NSFC) under Grant No. 61471268, the HongKong,
Macao and Taiwan Science & Technology Cooperation Program of China, and
Toshiba Research Europe Limited in the project “Massive MIMO Channel
Modeling and Applications”. The associate editor coordinating the review
of this letter and approving it for publication was Y. Shen. (Corresponding
author: Xuefeng Yin.)
J. Chen and X. Yin are with the College of Electronics and
Information Engineering, Tongji University, Shanghai 201804, China (e-mail:
09chenjiajing@tongji.edu.cn; yinxuefeng@tongji.edu.cn).
S. Wang is with the Telecommunications Research Laboratory,
Toshiba Research Europe Ltd., Bristol BS1 4ND, U.K. (e-mail:
stephen.wang@toshiba-trel.com).
Digital Object Identifier 10.1109/LCOMM.2016.2585478
widely used, the scatterer localization function can help direct
the antenna beam accurately to the scatterers that provide
significant contributions to the propagation channels.
Measurement-based channel models characterizing wave
propagation in various environments are important for algo-
rithm design and performance optimization. High-resolution
channel estimation algorithms, such as the Space-Alternating
Generalized Expectation-maximization (SAGE) algorithm [2],
and Richter’s Maximum likelihood estimation (RiMAX) algo-
rithm [3] make use of a generic multipath model to process
received signals under the planar wavefront assumption.
However, when it considers a large-scale antenna array struc-
ture, a spherical-wavefront-based channel estimation algorithm
may offer a more accurate description of the realistic channel.
The localization and characterization of scatterers are impor-
tant for channel modeling and establishment of map-based
environment. In [4], a spherical wave assumption based tri-
angulation principle is proposed, where the delay or angular
domain trajectories of MPCs observed by a moving Rx are
applied to detect the locations of scatterers. In [5], a spherical-
wavefront-based estimation is proposed to extract the distance
between the scatterer and the antenna array. A near-field source
localization method using the Direction of Arrival (DoA)
matrix is given in [6]. In this letter, the estimated distance
between the last-hop scatterer to a Rx reference antenna,
the azimuth of arrival (AoA) and elevation of arrival (EoA)
are used to determine the location of the last-hop scatterers.
Localization accuracy of last-hop scatterers is verified by
mapping the estimated scatterers to the real objects in a
measurement environment.
The rest of this letter is organized as follows. Section II
presents a spherical-wavefront-based generic model applied
to the cases with the architecture of a large-scale antenna
array. Section III proposes a scatterer localization algorithm
using the SAGE principle. Section IV elaborates the simulation
results to assess the proposed method. Section V shows
the measurement-based performance results, followed by a
conclusion in Section VI.
II. S
IGNAL MODEL
The planar wave assumption is valid when the scatterer
is further from the antenna array than the Rayleigh distance
d
Rayleigh
= 2D
2
/λ,whereD in meters is the aperture of the
array and λ represents the wavelength of the carrier frequency
[1, Ch. 2.2.3]. In the large-scale antenna array scenarios where
the array aperture is sufficiently large, the distance between
the array and a scatterer may be less than Rayleigh distance.
This makes the planar wavefront assumption inapplicable since
the wavefront observed at the locations of individual antennas
exhibits spherical curvature [5], [6].
1558-2558 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.