NLOS Identification and Mitigation for Geolocation Using Least-squares Support Vector Machines

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

20 Citations (Scopus)
341 Downloads (Pure)

Abstract

This paper examines the problem of non-line-of- sight (NLOS) identification and mitigation for geolocation signals in mobile networks. A ray tracing tool is used to simulate a mobile radio network with fixed base stations and thousands of mobile stations. The channel data between these mobile stations and base stations is used to extract parameters or features that are used for classification. Techniques for NLOS identification using a Least-Squares Support Vector Machine (LSSVM), are devised, producing greater than 98 percent accuracy for the proposed location specific approach, and 87 percent for the location independent approach. Respective LSSVM NLOS mitigation techniques are also proposed and evaluated. A usage context for the location specific approach is suggested, where the approach can help in addressing some of the challenges of next-generation wireless systems like massive MIMO.
Original languageEnglish
Title of host publicationWireless Networks of the IEEE WCNC 2017
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Number of pages6
ISBN (Electronic)9781509041831
ISBN (Print)9781509041848
DOIs
Publication statusPublished - 11 May 2017

Publication series

NameIEEE Wireless Communications and Networking Conference (WCNC)
PublisherIEEE
ISSN (Print)1558-2612

Fingerprint

Dive into the research topics of 'NLOS Identification and Mitigation for Geolocation Using Least-squares Support Vector Machines'. Together they form a unique fingerprint.

Cite this