A Probability-Based Acoustic Source Localization
Scheme Using Dual-Microphone Smartphones
Sanfeng Zhu, Naigao Jin, Xueshu Zheng, Han Yao, Shuailing Yang, Lei Wang
School of Software
Dalian University of Technology, China
sanfeng.zhu@gmail.com, ngjin@dlut.edu.cn, xueshuzheng.steve@gmail.com
yaohan125@gmail.com, yangsl1008@gmail.com, lei.wang@ieee.org
Abstract—This paper proposes a new acoustic source localiza-
tion scheme, called Probabilistic Cutting Method (PCM), with
randomly deployed smartphones which equipped with known
location and direction dual-microphones. Instead of using the
value of TDOA (Time Difference Of Arrival), we just use binary
information (0/1) and probability to convert the localization
problem into plane cutting issues. We can easily come up with
the Basic Cutting Method (BCM), but it may appear empty
set when error (location error, angle error or error anchors)
occurs. PCM can effectively avoid the problem along with lower
positioning error. When comparing PCM with TDOA and BCM
in different aspects, simulation evaluation results indicate that
PCM algorithm achieves highly robustness and accuracy.
Index Terms—Acoustic Source Localization, Binary Informa-
tion, Probability
I. INTRODUCTION
The growing scale and importance of computer technology
has driven the wide utilization of acoustic source localization
techniques in many occasions including locating speakers
in a room [1], shooter localization [2], network routing,
surveillance [3], target tracking [4], and emergency response.
Most existing localization technologies based on microphone
arrays usually have problems of low computing capacity, high
system costs, and low positioning accuracy in the presence of
reverberation environment.
The traditional microphone array-based localization systems
have been thoroughly studied in the literature. Despite their
obvious advantages over single-microphone systems, tradi-
tional microphone array-based localization systems have their
limitations because they usually sample the sound fields only
locally, typically at a relatively large distance from the sound
source(s). Furthermore, due to the constraints of space and
energy, especially in miniature and portable devices, the array
is often limited in physical size and processing power, what’s
more, it demands strict GCC (Generalized Cross-Correlation)
algorithm [5], [6].
Given these limitations, we believe there is an opportunity to
propose a time-synchronized method, in this paper, we explore
to locate a single acoustic source, and it makes three major
contributions, which can be described as follows:
Naigao Jin is the corresponding author in Dalian University of Technology.
(1) Synchronization free. We adopt smartphones which have
dual time-synchronized microphones to remove the time-
synchronized requirements of conventional localization
methods.
(2) Low costs, easy deployment. We implement a simple GCC
algorithm to weaken the demand of hardware to compute
the value of TDOA with randomly deployed smartphones,
by leveraging binary information and probability to locate
the target.
(3) High robustness and accuracy. Simulation results indicate
our algorithm has strong robustness and high accuracy
when location error, angle error and error anchors exist.
The rest of this paper is organized as follows: Related work
is reviewed in Section II. Section III presents the system
overview. Section IV states a simple method, but it may
involve empty set. In Section V, we introduce the detailed
design of PCM algorithm. The simulation of the system is
evaluated in Section VI. Finally, Section VII summarizes the
conclusions and future works.
II. R
ELATED WORK
Based on whether to calculate the distance between n-
odes, the current localization methods can be classified into
two kinds: range-based and range-free [7]–[9]. Range-based
method includes RSSI (Received signal strength indicator)
[10]–[12], TOA (Time of arrival) [13], [14], TDOA (Time
difference of arrival) [15], [16]. Range-free method can be
implemented by geometric methods, e.g., MDS (Multi dimen-
sional scaling) [17], DV-Hop (Distance vector-hop) [18].
Acoustic source localization is a well-studied problem. The
existing methods can be divided into four major categories:
RSS (Energy-based/received signal strength) [19], [20], AOA
(Angle of arrival) [21], TOA [22], and TDOA [16]. The RSS
method does not require time synchronization, however, it is
very sensitive to channel conditions. AOA method need to be
equipped with at least two microphones on each node, as well
as complicated microphone array processing techniques, and
demands high computational complexity. The TOA algorithm
employs the information of accurate signal transmission time
from source to the receiver; what’s more, it must pay attention
to time synchronization among participating nodes. Compared
1
IEEE ICC 2015 - Ad-hoc and Sensor Networking Symposium
978-1-4673-6432-4/15/$31.00 ©2015 IEEE 6621