![](https://csdnimg.cn/release/download_crawler_static/15405181/bg1.jpg)
Delay-Estimation-Based Asynchronous Particle Filtering for
Passive Target Tracking in Underwater Wireless Sensor
Networks
Meiqin Liu
1,2
, Lijia Zhao
2
, Senlin Zhang
2
1. State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou 310027, P. R. China
2. College of Electrical Engineering, Zhejiang University, Hangzhou 310027, P. R. China
E-mail: {liumeiqin, zhaolijia, slzhang}@zju.edu.cn
Abstract: Passive target tracking based on bearings-only or doppler-bearing measurements is difficult because there is no way
to determine the propagation time from source to receiver. This paper deals with the variable propagation delay problem in
3-dimensional passive tracking with the underwater wireless sensor networks (UWSNs). We propose an asynchronous particle
filtering algorithm based on delay estimation to overcome the difficulty that asynchronous data can’t be fused directly. Firstly,
according to the 3-d geometry structure among the snapshot state, truly measured state, and node position, we deduce the
particle delay from measured position to node position based on prior particle state. Then the measured state can be expressed
as a function of snapshot state and the likelihood of the measurements with prior snapshot states is derived. Thus we build the
relationship between snapshot state and asynchronous time-unclear measurements based on a reverse-transitivity particle filtering
algorithm. Simulation results show that the proposed algorithm is effective in complex 3-d scenes, and further experiments
illustrate that the fusion performance is related to both the target velocity and node number in UWSNs.
Key Words: Passive Target Tracking, Doppler-Bearing Measurements, Delay Estimation, Particle Filtering, Underwater Wire-
less Sensor Networks
1 Introduction
Underwater target detection is a significant problem in
military field. With the increasingly complex underwater
battlefield and the advanced technique of submarine, tradi-
tional underwater defense system in active mode has encoun-
tered many threats [1, 2]. To overcome the easy-detected
shortness of active detection, passive localization and track-
ing technology has been paid more and more attention. Pas-
sive tracking has many outstanding advantages such as self-
hiding, far detecting distance, and low energy consumption.
However with bearings-only or doppler-bearing measure-
ments of a single sensor, the target range cannot be obtained
directly, which makes the target unobservable and difficult
to track [3].
Several algorithms have been proposed to solve pas-
sive tracking problem. In Target Motion Analysis (TMA)
method, the sensor is deployed on a maneuvering platform,
then the target trajectory is observable and can be estimat-
ed using a Bearings-Only Tracking (BOT) algorithm [4–8].
The Doppler-Bearing Tracking (DBT) algorithm was further
developed for narrow band sonar applications [9, 10], which
can realize target localization even when the platform is not
maneuvering. Another method for target tracking is triangu-
lation technique in multiple sensors network, which calcu-
lates the target position at the triangulation points of bearing
lines [11, 12].
However, in underwater environment, the target speed is
not negligible compared to the speed of acoustic signals,
which makes the propagation delay so serious that the ap-
proaches mentioned above can’t be applied directly [13]. In
passive detection method, it is generally difficult to calculate
the delay, namely find out which instant that the snapshot
This work was supported by the National Natural Science Foundation
of China under Grants U1609204, 61374021, 61531015, and 61673345.
measurement really corresponds to. In many military appli-
cations, wave front curvature ranging algorithm is used to
localize the target by raw data analysis and time delay esti-
mation of multiple sensors [3, 14, 15], but sending raw data
to the fusion center requires much wide bandwidth and pow-
er, which is limited in distributed wireless sensor network
for large area surveillance [16, 17]. Direction of arrivals
(TOAs) is commonly used in underwater localization with
narrow-band and limited-energy scenes. With less bits of re-
quirement for communication, TOAs saves much energy and
can achieve similar performance as the earlier method [18].
In [19], Kaplana and Le deduced the retarded DOA in 2-d
scene and discussed when it is useful to employ propaga-
tion compensation (PC) to improve target localization, but
the 3-d environment hasn’t been considered. While in prac-
tical underwater environment, target tracking is commonly
executed in 3-d scenarios where the target motion is more
complex and the uncertainty of the target location greatly in-
creases compared to the 2-d scene. The compute method of
retarded DOA in [19] can’t be applied directly in 3-d scenes,
so it is essential to study the relationship between retarded-
DOA and snapshot-DOA and develop a new delay estima-
tion algorithm in 3-d scenarios.
Referring to K’s method, we adopt direction of arrival
(DOA) algorithm at each node and estimate the target posi-
tion by fusing DOAs from multiple nodes. The remainder of
this paper is organized as follows. In Section 2, the passive
tracking problem in the underwater wireless sensor network-
s (UWSNs) is formulated with the target model and sensor
model. In Section 3, we present the particle filtering algo-
rithm based on delay estimation to deal with asynchronous
data fusion problem. Simulation result is shown in Section 4,
and in Section 5 we discuss the performance of our methods
and present the conclusions.
Proceedings of the 36th Chinese Control Conference
Jul
26-28, 2017, Dalian, China
8929