NSF NETS: Radio Interferometric Tracking of Wireless Nodes Indoors

    Our focus in this NSF project (CNS 0721604) was to develop technologies for fine-grain localization of wireless nodes using radio interferometry and evaluate their applicability indoors.Our work is based on the original Radio Interferometric Positioning System (RIPS) introduced in 2005:

    M. Maroti, B. Kusy, G. Balogh, P. Volgyesi, K. Molnar, A. Nadas, S. Dora, and A. Ledeczi, "Radio Interferometric Geolocation", ACM Third International Conference on Embedded Networked Sensor Systems (SenSys 05) 

     TripLoc

    TripLoc is a novel localization and tracking method that relies on basic RIPS measurements, but overcomes many of its limitations. The basic idea is to group together three of the four nodes involved to form an antenna array and act as an anchor node. The two transmitters and one of the receivers are arranged in such a manner that their antennas are mutually orthogonal to eliminate parasitic antenna effects. Since all the antennas are within half the wavelength, the modulo ambiguity is also eliminated. Each target node that needs to be localized, acts as the second receiver in this scheme. The measured phase difference between a target node and the receiver of the anchor node array constrains the location of the node to a hyperbola. The asymptote of the hyperbola can be considered a bearing estimate assuming that the node is not too close to the array.

    To the best of our knowledge, this is the first system that is able to estimate RF bearing using simple motes and no additional hardware whatsoever. For more information, read this paper:

    I. Amundson, J. Sallai, X. Koutsoukos, and A. Ledeczi, "Radio Interferometric Angle of Arrival Estimation", 7th European Conference on Wireless Sensor Networks (EWSN), 2010.

     Reverse Quasi Doppler Localization

    It is well known that the bearing to a radio source can be estimated by an array of antennas typically arranged in a circular manner. The method is often referred to as Quasi-Doppler measurement. The disadvantage of the existing method is that the receiver is relatively large because of the multiple antennas (typically 8 or 16) and it is computationally intensive to process the high frequency radio signals. Thus, it cannot be done on small, inexpensive radio tags. Instead, we propose to use the array on the transmitter side utilizing as few as three antennas. That is, we switched the roles of the receiver and transmitter, hence the name, Reverse Quasi Doppler measurement. 

      The interferometric technique is utilized to transform the useful phase information from the high frequency radio signal to a low frequency signal (< 1 kHz) that can be processed on low-cost hardware. Utilizing three anchors nodes with small antenna arrays, any number of low cost wireless nodes with single antennas can be accurately localized. The advantage of this Quasi Doppler approach over the original RIPS is its speed and scalability. Localization takes only a few seconds as opposed to several minutes and the time does not depend on the number of nodes. 

    To the best of our knowledge, we are the first to suggest this simple technique that enables low-cost wireless nodes to estimate their bearing. Our paper describing the technique was presented at the IPSN conference in 2009:

    J. Sallai, P. Volgyesi, and A. Ledeczi, "Radio Interferometric Quasi Doppler Bearing Estimation", Information Processing in Sensor Networks (IPSN 09, SPOTS Track)

     Doppler Shift-based Tracking and Navigation

    We extended RIPS to measure the Doppler shifts caused by moving nodes to estimate their velocity and location simultaneously. Consequently, the new approach allows for simpler and faster tracking than the original RIPS technique:

    1. Each of the receivers measures the Doppler shift only, whereas the relative phase measurements needed to be carried out by pairs of receivers, and
    2. Doppler shifts are measured at a single carrier frequency, whereas up to 21 different frequencies were required to obtain accurate ranging data from phase measurements.

    The new results indicate that COTS Mica2 motes can measure RF Doppler shifts with 0.2 Hz accuracy corresponding to a 0.14 m/s error in relative speed estimates using radio interferometric technique.

    The tracking problem is modeled as a non-linear optimization problem and an extended Kalman filter is used to solve it accurately assuming Gaussian measurement errors. However, this approach may fail if the tracked node changes its speed or direction. Instead we update the Kalman filter state by performing constrained least-squares optimization when a maneuver is detected. The combined approach achieves almost a 50% accuracy improvement over the Kalman filter alone when the mobile node changes its direction and speed frequently.

    This work was presented at Sensys 2007 where it received the Best Paper Award:  

    B. Kusy, A. Ledeczi, and X. Koutsoukos, "Tracking mobile nodes using RF Doppler shifts", Proceedings of the 5th International Conference on Embedded Networked Sensor Systems (SenSys '07)

    Publications

    Click on the title above to see most of the publications resulting from this project.

    Downloads

    The following software packages are available in open source format: