Location Estimation via Support Vector Regression

Wu, Zhen; Li, Chun-hung; Ng, Joseph Kee‐Yin; Leung, Kwok Wa

doi:10.1109/tmc.2007.42

Cited by 177 publications

(97 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, as will be shown later, this approach of handling missing MI values has proved to be more robust compared to Wu et al (2007). In the case of the SVC&SVR models, 95% DE is influenced additionally by the cases where the subarea is incorrectly selected.…”

Section: S V M -B a S E D M O D E L S Fo R Po S I T I O N I N G Nomentioning

confidence: 98%

“…To credibly verify the performances of the proposed SVR, SVC&SVR 2 × 2 and SVC&SVR 4 × 4 models, we have applied the exact same field data to some of the most promising techniques from Section 2: SVR with SoE kernel function -SVR SoE (Wu et al, 2007), DCM (Laitinen et al, 2001), ANN (Takenga et al, 2006) and EKF trained ANN (Anne et al, 2004). SVR SoE was chosen in order to credibly verify the proposed approach for handling missing RSS values in the database.…”

Section: S V M -B a S E D M O D E L S Fo R Po S I T I O N I N G Nomentioning

confidence: 99%

“…In Wu et al (2007), positioning models were implemented using field measurements. For each measurement location, its fingerprint was taken and stored in a database, which was used for model training and verification.…”

Section: Introductionmentioning

confidence: 99%

“…In the first model, a SVR algorithm was applied. In contrast to Wu et al (2007), we propose the use of a Laplace kernel function where missing RSS values in database fingerprints are replaced with a fixed value below the receiver threshold, which was later shown to provide better performances. Moreover, a quadratic ε-insensitive loss function was used in the optimisation criterion in order to suppress large output errors.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SVM-based Models for Mobile Users' Initial Position Determination

Petric

Nešković

et al. 2014

J. Navigation

View full text Add to dashboard Cite

A large interest in developing commercial Location-Based Services (LBS) and the necessity of implementing emergency call services, have led to the intensive development of techniques for mobile users' localisation. In this paper, a Public Land Mobile Networks (PLMN) -based technique for initial position determination is proposed as an alternative to satellite-based methods in environments with obstructed satellite signals. Two positioning models, based on handset available Received Signal Strength (RSS) measurements from Global System for Mobile Communications (GSM) base stations and the use of Support Vector Machine (SVM) algorithms, are proposed. Performances of proposed models are verified using field measurements, collected in a suburban environment. Models are analysed in terms of positioning accuracy, complexity and latency, and compared to some other promising PLMN-based techniques. Using proposed SVM-based positioning models a median error of 4·3 m-6·2 m and latency of less than a second can be achieved. K E Y

show abstract

Section: S V M -B a S E D M O D E L S Fo R Po S I T I O N I N G Nomentioning

confidence: 98%

Section: S V M -B a S E D M O D E L S Fo R Po S I T I O N I N G Nomentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SVM-based Models for Mobile Users' Initial Position Determination

Petric

Nešković

et al. 2014

J. Navigation

View full text Add to dashboard Cite

show abstract

“…Because using RSSI as a measurement is simple and does not require special hardware several indoor positioning methods make use of it. There are plenty examples such as, Horus which makes use of position estimation by probability including RSSI information [4]. There are two main problems with RSSI based methods.…”

Section: Introductionmentioning

confidence: 99%

An Indoor Positioning Method using IEEE 802.11 Channel State Information

Escudero¹,

Hwang²,

Park³

2017

Journal of Electrical Engineering and Technology

View full text Add to dashboard Cite

-In this paper, we propose an indoor positioning system that makes use of the attenuation model for IEEE 802.11 Channel State Information (CSI) in order to determine its distance from an Access Point (AP) at a fixed position. With the use of CSI, we can mitigate the problems present in the use of Received Signal Strength Indicator (RSSI) data and increase the accuracy of the estimated mobile device's location. For the experiments we performed, we made use of the Intel 5300 Series Network Interface Card (NIC) in order to receive the channel frequency response. The Intel 5300 NIC differs from its counterparts in that it can obtain not only the RSSI but also the CSI between an access point and a mobile device. We can obtain the signal strengths and phases from subcarriers of a system which in turn means making use of this data in the estimation of a mobile device's position.

show abstract

A passive location system for single frequency networks using digital terrestrial TV signals

Yan

2011

Trans Emerging Tel Tech

View full text Add to dashboard Cite

In this work, a passive location system for single frequency networks using digital terrestrial signals is proposed. A crosscorrelation method that requires a correlation with fixed frequency was introduced for time delay estimation. The idea is first online reconstruction and update of the transmitted copy of the nearest base station (BS), and then uses this signal copy for correlation with the received signal samples. A time delay extraction method combined with K-mean clustering algorithm is developed. The minimum value of each cluster is extracted as the time delay of each BS. The BS identification can be modelled as a data association problem. A new gate that requires little statistical knowledge was utilised to solve time-of-arrival measurement uncertainty. Meanwhile, the data association strategy is also developed for successive data stream. Mobile station location is formulated as an optimisation problem and solved by a search processing. Simulations study was conducted to evaluate the system performance for different scenarios. The results demonstrate its robustness and high performance.

show abstract

Location Estimation via Support Vector Regression

Cited by 177 publications

References 9 publications

SVM-based Models for Mobile Users' Initial Position Determination

SVM-based Models for Mobile Users' Initial Position Determination

An Indoor Positioning Method using IEEE 802.11 Channel State Information

A passive location system for single frequency networks using digital terrestrial TV signals

Contact Info

Product

Resources

About