|M.Sc Student||Evgeny Kupershtein|
|Subject||Indoor Localization based on Multipath Fingerprinting|
|Department||Department of Electrical Engineering||Supervisors||Full Professor Cohen Israel|
|Dr. Matti Wax|
|Full Thesis text|
In recent years there has been a growing interest in position location in indoor venues. As more applications requiring indoor localization are emerging in the market, the demand for accurate and reliable localization increases. Unfortunately, the accuracy of available techniques is limited, and a dense and expensive deployment is required. The problem of accurate indoor localization is challenging due to severe multipath conditions existing in indoor environments. As a result, the classical position location techniques based on the line-of-sight (LOS) condition are not valid in such scenarios.
This research presents a novel method enabling single-site localization of wireless emitters in a rich multipath environment. The localization is based on a novel fingerprinting technique exploiting the spatial-temporal characteristics of the multipath signals received by the base station antenna array. The fingerprint is based on a lower dimensional signal subspace of the spatial-temporal covariance matrix, capturing the dominant multipath signals. The subspace approach does not require estimation of the directions-of-arrival and differential-delays of the multipath reflections, which is both difficult and computationally intensive problem in rich multipath environments. The proposed method exploits fingerprint matching based on the powerful similarity-profile criterion with effective complexity reduction techniques limiting the matching complexity and providing storage saving.
The proposed fingerprinting technique is investigated in the time and frequency domains showing a similar level of accuracy. Both approaches are applicable to most modern communication techniques and do not require new hardware on the user device.
Necessary and sufficient conditions that guarantee unique localization are presented. The performance is validated with both simulated and real data, demonstrating localization accuracy of about 1m in typical indoor environments.