Applying a self-learning fingerprinting approach to enhance the RTT multilateration positioning system in IEEE 802.11mc

Description

Nowadays, indoor positioning is a still an open issue. The main reason, though not the only one, is that location indoors require accurate measurements, i.e. positioning errors of 1 to 2 meters in average. There are technologies, such as those based in ultra wideband (UWB) that are able to provide positions with only centimeters of uncertainty. However, those technologies require custom network and user equipment to work, which often make them costly and hard to deploy. 

Lots of efforts have been addressed to use communication networks for location purposes. However, the problem then is that positioning errors rise (noticeably) over the 2 meters.

In the recent past, the Task Group mc (TGmc) of the IEEE 802.11 Working Group

(also referenced as the IEEE 802.11mc) proposed some enhancements to the 802.11 protocol so that it provided accurate time-of-flight measurements suitable for indoor positioning. This amendmend was introduced in the 2016 revision of the protocol, but it haven’t been adopted by main manufacturers until now. Recently, Google have dramatically supported this approach by allowing fine RTT measurements in any smartphone running Android 9.0. Other main actors in the WiFi field (such as Intel or Cisco) have implemented Fine Time Measurements in their WiFi cards and access points as well. 

RTT location in IEEE 802.11mc is proposed as an active multilateration ranging-based solution. This means that location traffic needs to be injected (FTM request and responses) so that the RTT can be properly computed. Active location solutions are known to be not scalable (the larger the amount of stations being located, the lower throughput available for data services). 

RSSI fingerprinting is a well-known passive solution for indoor positioning, based on building a signal strength map at known reference emplacements (i.e. the fingerprinting database), which is further used to compare live measurements to those previously recorded and infer the right location the station is.

This project is aimed at coupling these two approaches, active RTT multilateration with passive RSSI fingerprinting, without the requirement of a previously built database.

Objectives

The goal of this project is to couple these two approaches, active RTT multilateration with passive RSSI fingerprinting, without the requirement of a previously built database. This goal would involve the following stages:

1. Provide the location system specification. This needs to fulfill the following requirements:

• Zero knowledge fingerprinting database.
• Self-learning algorithm.
• High scalability.
• Consistent accuracy, either the active, passive or hybrid mode is used.

2. Implement a simulation model for the system specified.
3. Analyzing the performance of the new location system.

Period

Start date: --
End date: --
Status: Open

Funding

No funding is provided for this project.

Participants

Supervisor: Israel Martin Escalona (GRXCA)

Student: To be assigned.

Further information

No additional information is provided.