by Francisco Barcelo-Arroyo, Israel Martin-Escalona and Marc Ciurana-Adell
The ability to pinpoint a terminal’s position is useful for many applications of mobile computing and for network optimization (eg handovers, tariffs, resource management). A range of techniques are available to obtain a terminal’s position [1]. GPS, for example, is used externally to the network and achieves good accuracy outdoors, with the trade off of increased energy consumption. Communication devices, however, are frequently used indoors, connecting to private networks, such as WLAN. Since GPS is inaccurate indoors owing to signal blockage and multipath errors, further research on indoors localization through communication networks is required. Mobile computing is linked to indoors positioning in applications such as: aged care, remote health control and security of buildings such as hospitals.
The fingerprinting technique is used extensively for WLAN positioning. The terminal collects the received signal strength from several access points and, during a precalibration phase, compares the achieved vector to the vectors previously recorded along with their positions. This technique does not involve modifications to the hardware. Other techniques use the time of flight (ie the time needed by a signal to travel between two nodes) to estimate the distances to several access points at known positions and then apply a trilateration process. The time of flight is more consistent than the signal strength. But in order to avoid modifications to the terminal’s hardware, the time of flight must be obtained from communication messages by using only software.
Recent research at the Technical University of Catalonia has led to a procedure to measure distances between terminals [2]. This procedure is aimed at obtaining the time of flight after adding timestamps to messages sent and to the corresponding acknowledgements received. The round trip time (RTT) is computed as the difference between both timestamps, and the distance between the nodes is inferred by considering that the trip occurs at the speed of the radio signal. The software must interact with the link layer of the protocol stack of the device. A simple approach is to use a network interface capable of providing time measurements made by the hardware, but the timestamps performed do not have an acceptable accuracy for many location applications (eg the characteristics of IEEE 802.11 lead to a resolution of one microsecond corresponding to 300 metres in distance.)
A more sophisticated approach is presented in Figure 1. The protocol stack of the terminal is updated by introducing two software layers that are registered in each terminal. The registration process is run once and replaces the network interrupt handler (responsible for handling the events related with the network interface) with a new one. This new handler analyses the traffic transmitted and received and adds timestamps to certain messages so that the RTT can be finally computed. Figure 1 also shows the network architecture including the interrupt handler in a Linux-based device. The sources of the Linux kernel have been patched to allow location metrics (ie RTT) to be observed. These changes alter the mac80211 subsystem, which implements most of the common MAC features in Linux. The goals of these changes are 1) to allow the location-related capabilities to be registered and released and 2) to add timestamps to the messages exchanged between the terminal and the access point. The capabilities are implemented as plugins, so that each works as standalone. This allows isolation of the bugs and extension of the capabilities without impacting those that are already working. An RTT plugin has been developed in order to calculate the RTT between a node and an IEEE 802.11 access point. This plugin is responsible for most of the tasks developed by the interrupt handler. Specifically, it filters the traffic not suitable for location purposes and matches the transmission and reception messages involved in an RTT, so that the RTT can finally be computed. The interaction between the user’s applications and the RTT plugin is done by means of system calls to a new module named pos80211 [3]. This module provides the computed and buffered RTTs to the user’s applications.
Currently, the proposed implementation has been prototyped and tested at distances shorter than 30 metres with good results. Future work includes more testing (eg different scenarios, longer distances) and developing new plugins for hyperbolic, instead of circular, trilateration. The current plugin provides the RTT which is useful for circular trilateration, while time differences are appropriate for hyperbolic. The use of this proposal in ad hoc networks is also investigated.
References:
[1] Y. Gu, A. Lo, I. Niemegeers: “A survey of indoor positioning systems for wireless personal networks”, IEEE Commun. Surveys & Tutorials, vol. 11, no. 1, pp. 13-32, 1Q 2009
[2] F. Barcelo-Arroyo, M. Ciurana, I. Martin-Escalona: “Process and system for calculating distances between wireless nodes”, U.S. Patent 8 289 963, October 26, 2012
[3] J. Corbet, A. Rubini, G. Kroah-Hartman: “Linux Device Drivers”, O'Reilly Media, Third edition, 2005
Link: http://grxca.upc.edu
Please contact:
Francisco Barcelo-Arroyo, Universitat Politecnica de Catalunya, Spain
Tel: +34 934016010
E-mail:
