Wi-Fi-AWACS: "The Ad hoc Wireless locAtion and traCking System" Project.

The goal of the Wi-Fi-AWACS project is to develop a software suite for indoor and outdoor 3D location and tracking the Wi-Fi nodes using distributed, mobile, low density grid of heterogeneous sensors.

A geolocation algorithm used in the system is described in the article draft "Reduction of Deviation Selective Algorithms for the Wireless Location and Tracking System" .

If you have questions, you are welcome to write to: selitsky@yahoo.com

Hardware and Software requirements:

• The system collects data passively. Therefore, a Wireless NIC driver, capable to support monitor mode, should be available for your hardware and OS (Linux, BSD (see wi and ath), Mac OS X).
• The system uses pcap library to get raw packets from the kernel space. Therefore you need to get or compile pcap library for your OS.

The full-scale functionality will be implemented incrementally. Here is the list of functional features by a release:

• wifi-awacs.0.1a.tgz Added rod utility - implementation of fast ROD filtering algorithm to find 3 best anchors an triangulate using them.
• wifi-awacs.0.0l.tgz Major cleanup.
• wifi-awacs.0.0k.tgz New utility brute - brute force search for optimal 3 anchors for triangulation by minimization of predicted located error.
• wifi-awacs.0.0j.tgz Radius calculates R for Sr=S0-20logR-10kRlog(e) attenuation function, too. Dgraphgen plots real and expected R error maps.
• wifi-awacs.0.0i.tgz Sniffer sets transmitter's mac and coordinates for old data, too. Bug fix in Calibr for Sr=S0-10nlogR coefficent calculation. Added polynomial coefficient calculation. Dgraphgen plots radio maps of signal strength and error between actual and predicted signal strength for 3 analytical functions used in Calibr.
• wifi-awacs.0.0h.tgz Bug fixes and a new dgraphgen utility. Run dgraphgen < with --sigma or --power options to create density diagrams of power deviation and power on receiver itself to feed in webMatematica plotter. For example via http://cose.math.bas.bg/webMathematica/MSP/Sci_Visualization/Contour3DDataPlot interface.
• wifi-awacs.0.0g.tgz - This particular release was used on WRTSL54GS Wi-Fi router running OpenWrt linux distribution. It does not have an open source wireless driver, but Broadcom binary driver supplied with it supports monitor mode and conveniently reports frame headers in wlan_ng format. Another advantage is that Broadcom driver creates additional prism0 interface to allow access to monitor data, at the same time keeping main eth2 interface in operational mode. This, and also size, price and USB port makes the device an excellent choice for brains for a mobile robot.

  Visit OpenWrt site for instructions on how set crossplatform of native development environment: http://wiki.openwrt.org/BuildingPackagesHowTo

  With the current 0.0.g release instead of manual coordinate input, you may run sniffer with --xr , --yr and --zr options to set a receiver coordinates. Options --mact , --xt, --yt, --zt set transmitter coordinates for particular MAC.
• wifi-awacs.0.0f.tar.gz - clean up and fixes
• wifi-awacs.0.0e.tar.gz - proof of concept
  • The coordinate system is local rectangular. Coordinate input is manual.
  • The only local receiver data are used. There is no information sharing among the sensor nodes.
  • The system determines locations for the static Wi-Fi nodes.
  • The receivers and transmitters are assumed to be geometrically homogeneous (in terms of transmit power and receiver sensitivity).
  • The environment is isotropic, an attenuation factor is determined empirically during the calibration process.
  • This is a comand string batch system. The sniffer, calibrate and triangulate processes work sequentially.
  • The sniffer works with Linux wlan_ng driver frame headers. I.e. Prizm and ACX100 chipsets are supported.
  • There is no prior filtering of the data before the triangulation.
  • The only one attenuation approximation is used: S=S0-10nlogR.

Other Projects