Indoor localization using unsupervised manifold alignment with geometry perturbation

Khaqan Majeed, Sameh Sorour, Tareq Y. Al-Naffouri, Shahrokh Valaee

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Scopus citations


The main limitation of deploying/updating Received Signal Strength (RSS) based indoor localization is the construction of fingerprinted radio map, which is quite a hectic and time-consuming process especially when the indoor area is enormous and/or dynamic. Different approaches have been undertaken to reduce such deployment/update efforts, but the performance degrades when the fingerprinting load is reduced below a certain level. In this paper, we propose an indoor localization scheme that requires as low as 1% fingerprinting load. This scheme employs unsupervised manifold alignment that takes crowd sourced RSS readings and localization requests as source data set and the environment's plan coordinates as destination data set. The 1% fingerprinting load is only used to perturb the local geometries in the destination data set. Our proposed algorithm was shown to achieve less than 5 m mean localization error with 1% fingerprinting load and a limited number of crowd sourced readings, when other learning based localization schemes pass the 10 m mean error with the same information.
Original languageEnglish (US)
Title of host publication2014 IEEE Wireless Communications and Networking Conference (WCNC)
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Number of pages6
ISBN (Print)9781479930838
StatePublished - Apr 2014

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01


Dive into the research topics of 'Indoor localization using unsupervised manifold alignment with geometry perturbation'. Together they form a unique fingerprint.

Cite this