A novel location estimation based on pattern matching algorithm in underwater environments:
abstract:
In this paper, we present a novel approach based on pattern recognition to treat the underwater localization. The goal is to achieve underwater localization by the pattern matching algorithm. It should be noted that the reflected signals in underwater environments do not affect our location estimation. Therefore, the underwater localization in this study is straightforward and efficient by using the pattern matching algorithm. We exploit the maximum likelihood (ML) to perform our study. Initially, the underwater signals are collected by the sound receiver at some sampling locations. These signals are suitably processed by the ML models and are stored in database. The test location in real-time is estimated through the database. Experimental results show that good accuracy of positioning can be obtained by proposed schemes. The proposed localization schemes can be applied to many other applications in underwater environments.
1. Introduction
Source localization plays an important role in radar, sonar, geological exploration, and wireless communication. In general, it can involve the joint estimation of frequencies, Doppler shifts, directions of arrival (DOA) and time/time difference of arrival. The geometric approaches such as joint estimation of DOA and time delay are based on transmitting a known signal and estimating DOA from time delay of the reflected signal [1–5]. Although the above approaches show good performance, it is usually difficult to deal with the reflected signals and is then affected by multi-path fading effect.
This motivates us to develop an alternative approach for underwater localization.
In this paper, we present a novel approach for underwater localization based on WLAN (wireless local area network) location fingerprinting [6–8]. The fingerprinting based localization behaves
like the process of pattern matching, which is divided into two stages, i.e., the offline and online stages. During the offline stage, signals collected by the sound receiver at the sampling locations
are stored to constitute the database. During the online stage, the signal in real-time is measured and the location can be estimated based on the database built in the offline stage.
The aim of our study is to achieve underwater localization by the pattern matching algorithm. Note that the reflected signals do not affect our location estimation. The underwater localization in this study is straightforward and efficient because the pattern matching algorithm does not require pure direct-radiation signals.
Experimental results show that the good accuracy of positioning is obtained by proposed schemes. Therefore, the innovation of underwater localization can be accomplished based on our approach.
In Section 2, the theoretical formulations are given. Experimental setup and results are given in Section 3. Finally, the conclusion is given in Section 4
2. Formulations
2.1. Preparation for signal collection
For transmitting and receiving signals in the underwater environment, in this study a fish finder is adopted as the sound projector to transmit the AM (amplitude modulation) signals and a hydrophone is adopted as the sound receiver. During the offline stage, the AM signals in the underwater environment are collected by the sound receiver at M reference locations.We then convert these AM signals from time-domain to frequency-domain through Fast Fourier Transform (FFT) and the magnitudes of spectrum at the specific frequency from each sound projector are stored to constitute the database. During the online stage, the actual location coordinate of the sound receiver can then be estimated based on the database built in the offline stage by pattern matching algorithms.
3. Results
3.1. Experimental setup
In this section, we begin with a description of our experimental setup and experimental results. We perform our experiment in the towing tank of the Systems and Naval Mechatronic Engineering
Department of National Cheng Kung University, Tainan, Taiwan, as shown in Fig. 1. The dimension of the towing tank is 175 m(L) 8 m(W) 4 m(D) meters. The experimental setup in the towing tank is shown in Fig. 2. We adopt two GARMIN-340C fish finders as the physical sound projectors and one B&K 8106 hydrophone as the sound receiver. Both of the two fish finders transmit AM signals. The center frequencies of the two transmitters are 50 kHz and 200 kHz, respectively. The procedures of signal collection stage are shown in Fig. 3. An NI-6251 DAQ (data acquisition) card is utilized to gather the AM signal from each sound projector. In Fig. 3, the notebook PC is installed with the LabVIEW-8.2 software for data acquisition and instrument control. The received signals are converted from time-domain to frequency-domain through FFT. Note that the above signal acquisition and processing are implemented by the commercial software LabVIEW-8.2.
In that software, the length of FFT is 262144, the type of time window is Hanning, and the sampling frequency is chosen as 500 kHz. We collect 100 samples of spectrum magnitudes at 84 (M = 84) reference locations separated by 1 m, as shown in Fig. 2.
Then we adopt the same procedure [13] to partition the data set into 2 independent groups, 10 samples for testing and 90 samples for training. In the parametric training, we estimate the means and
standard deviations of the likelihood functions from the training data and these statistical parameters are stored in the database.
DOWNLOAD LINK:
CLICK ME
abstract:
In this paper, we present a novel approach based on pattern recognition to treat the underwater localization. The goal is to achieve underwater localization by the pattern matching algorithm. It should be noted that the reflected signals in underwater environments do not affect our location estimation. Therefore, the underwater localization in this study is straightforward and efficient by using the pattern matching algorithm. We exploit the maximum likelihood (ML) to perform our study. Initially, the underwater signals are collected by the sound receiver at some sampling locations. These signals are suitably processed by the ML models and are stored in database. The test location in real-time is estimated through the database. Experimental results show that good accuracy of positioning can be obtained by proposed schemes. The proposed localization schemes can be applied to many other applications in underwater environments.
1. Introduction
Source localization plays an important role in radar, sonar, geological exploration, and wireless communication. In general, it can involve the joint estimation of frequencies, Doppler shifts, directions of arrival (DOA) and time/time difference of arrival. The geometric approaches such as joint estimation of DOA and time delay are based on transmitting a known signal and estimating DOA from time delay of the reflected signal [1–5]. Although the above approaches show good performance, it is usually difficult to deal with the reflected signals and is then affected by multi-path fading effect.
This motivates us to develop an alternative approach for underwater localization.
In this paper, we present a novel approach for underwater localization based on WLAN (wireless local area network) location fingerprinting [6–8]. The fingerprinting based localization behaves
like the process of pattern matching, which is divided into two stages, i.e., the offline and online stages. During the offline stage, signals collected by the sound receiver at the sampling locations
are stored to constitute the database. During the online stage, the signal in real-time is measured and the location can be estimated based on the database built in the offline stage.
The aim of our study is to achieve underwater localization by the pattern matching algorithm. Note that the reflected signals do not affect our location estimation. The underwater localization in this study is straightforward and efficient because the pattern matching algorithm does not require pure direct-radiation signals.
Experimental results show that the good accuracy of positioning is obtained by proposed schemes. Therefore, the innovation of underwater localization can be accomplished based on our approach.
In Section 2, the theoretical formulations are given. Experimental setup and results are given in Section 3. Finally, the conclusion is given in Section 4
2. Formulations
2.1. Preparation for signal collection
For transmitting and receiving signals in the underwater environment, in this study a fish finder is adopted as the sound projector to transmit the AM (amplitude modulation) signals and a hydrophone is adopted as the sound receiver. During the offline stage, the AM signals in the underwater environment are collected by the sound receiver at M reference locations.We then convert these AM signals from time-domain to frequency-domain through Fast Fourier Transform (FFT) and the magnitudes of spectrum at the specific frequency from each sound projector are stored to constitute the database. During the online stage, the actual location coordinate of the sound receiver can then be estimated based on the database built in the offline stage by pattern matching algorithms.
3. Results
3.1. Experimental setup
In this section, we begin with a description of our experimental setup and experimental results. We perform our experiment in the towing tank of the Systems and Naval Mechatronic Engineering
Department of National Cheng Kung University, Tainan, Taiwan, as shown in Fig. 1. The dimension of the towing tank is 175 m(L) 8 m(W) 4 m(D) meters. The experimental setup in the towing tank is shown in Fig. 2. We adopt two GARMIN-340C fish finders as the physical sound projectors and one B&K 8106 hydrophone as the sound receiver. Both of the two fish finders transmit AM signals. The center frequencies of the two transmitters are 50 kHz and 200 kHz, respectively. The procedures of signal collection stage are shown in Fig. 3. An NI-6251 DAQ (data acquisition) card is utilized to gather the AM signal from each sound projector. In Fig. 3, the notebook PC is installed with the LabVIEW-8.2 software for data acquisition and instrument control. The received signals are converted from time-domain to frequency-domain through FFT. Note that the above signal acquisition and processing are implemented by the commercial software LabVIEW-8.2.
In that software, the length of FFT is 262144, the type of time window is Hanning, and the sampling frequency is chosen as 500 kHz. We collect 100 samples of spectrum magnitudes at 84 (M = 84) reference locations separated by 1 m, as shown in Fig. 2.
Then we adopt the same procedure [13] to partition the data set into 2 independent groups, 10 samples for testing and 90 samples for training. In the parametric training, we estimate the means and
standard deviations of the likelihood functions from the training data and these statistical parameters are stored in the database.
DOWNLOAD LINK:
CLICK ME
No comments:
Post a Comment