OwlPS experiments at the CANMET mine, 2013
These experiments were conducted at the level -70 m of the CANMET laboratory mine near Val-d'Or, QC, Canada, in June 2013. The main goal was to test the behaviour of Wi-Fi positioning in underground mine tunnels.
Except if stated otherwise, all the scenarios follow these rules (or should follow them for future tests). The description of the scenarios has precedence over these common rules. Moreover, the report files associated with each test should also warn about each noticed mistake, and each exception made to these rules or to the scenario description.
These measurements took place at the level -70 metres of the CANMET laboratory mine, which is pictured on the plan ./figures/canmet70.svg.
CANMET is a formerly exploited gold mine. At the level -70 m, the tunnels are between about 2.50 and 3-metre wide, with some exceptions such as larger rooms or recesses. The walls are very irregular, and the actual width can vary by several dozens of centimetres depending on where it is measured.
The drifts are fairly long, the three main sections used in this series of experiments being around 130 metres long. They are relatively straight, but turn enough that it is easy to loose the line of sight between two devices that are a few dozens of metres apart.
In some tunnels, "doors" made of plastic blades hanging from the ceiling are installed; they are represented by short grey lines on the map. The short black line at the south of CP3 is a wooden wall with a door and a plastic tarp window, that closes a small heated room used an office when experimenting at this level of the mine.
Each tunnel section contains at least one water pipe (diameter 6 or 12 cm) and one compressed air pipe (diameter 6 cm). The southern tunnel contains a second 6-cm pipe which is empty. All these pipes are metallic and are attached to one of the walls of the drift. Additionally, an important group of cables and wires (network, electricity, radio, etc.) is attached to the ceiling.
All the wireless devices (mobile terminal and capture points) are equipped with Mini PCI MikroTik RouterBoard R52Hn IEEE 802.11a/b/g/n cards. Since we work only with 802.11b/g, only one of the two MC connectors of the card is used (see the next paragraph, Antennas). The Linux driver for these cards is ath9k. The power transmission used is 25 dBm.
All the wireless devices are equipped with an RP-SMA triband 2.4/5.1/5.8 GHz rubber ducky antenna, connected to the device with a RP-SMA to Type N adapter and a 22 cm long cable with a MC connector plugged into the wireless card. The gain of these antennas at 2.4 GHz is 3 dBi.
The mobile terminal is a MikroTik RouterBoard 411AH powered by a portable car battery booster kit that provides 110 V AC power plugs (and is freaking heavy!). The operating system is OpenWrt 12.09.
Like the mobile terminal, they all run OpenWrt 12.09.
Lenovo ThinkPad X200 running Debian GNU/Linux unstable (Linux 3.9).
CP1, CP2 and CP5 are powered by a 24 V PoE (Power over Ethernet) switch located next to CP1; CP3 and CP6 are powered by two PoE injectors located next to CP3.
The capture points' coordinates are also given in the OwlPS Positioner's configuration file ./owlps-config/listeners-rb.csv, and you can also visualise their positions on the map ./figures/canmet70.svg.
The capture points and the aggregation server are connected through an Ethernet network. The deployment plan ./figures/canmet_deployment.svg shows the wiring used, that takes advantage of the fact than the RouterBoard 433 have 3 Ethernet connectors (the 3 ports of each board are set up in a bridge). The IP network used in the wired deployment is 192.168.88.0/24. Each capture point has an address of the form 192.168.88.10x, where x is the capture point's number (CPx). The aggregation server has address 192.168.88.254.
The IP network used in this ad hoc network is 192.168.89.0/24; the addresses are equivalent to the addresses in the wired network (192.168.89.10x).
The mobile terminal is connected to a portable Wi-Fi access point (AP), which is a smartphone running CyanogenMod 10.1.0. The positioning requests are transmitted to the AP's IP address. Though it would have been possible to use the ad hoc network created by the capture points for this purpose, this would have implied to set up routing between the capture points (to avoid having to change the destination host from one capture point to another) and seemed a unnecessary complication. Moreover, it allows to control the transmission of the positioning requests simply by turning on and off the AP.
The temperature and humidity could not be measured precisely during the experiments. However, in the summer, it is known that the temperature is between 5 and 7 °C, and the humidity is very high.
OwlPS Aggregator version v1.3.3-72-g4d8e9cf was used (slightly earlier versions were also used in the first days of the experiments).
OwlPS Aggregator is run with the default parameters as of the version used, with autocalibration enabled. These parameters can be found in the configuration file ./owlps-config/owlps-aggregator.conf.
These parameters are not very important, except for the delay between two autocalibration orders. The default value is 1000 ms.
OwlPS Listener version v1.3.3-17-gdc60892 was used.
The OwlPS Listener program runs continuously, with the autocalibration activated. The exact parameters can be found in the configuration file ./owlps-config/owlps-listener.conf.
OwlPS Client version v1.3.3-17-gdc60892 was used.
The complete command used to launch OwlPS Client is the following:
owlps-client -i 192.168.43.1 -n20 -t10 -F800
For simplicity's sake, we defined a bunch of measurement points that can be used in the description of the scenarios. They are named after their proximity to the capture points.
In these scenarios, the client terminal is static, on a wooden table (cf. supra, Measurement-related rules). The measurements are taken for at least 5 minutes.
The client terminal is located at the emergency exit (MP61).
The client terminal is located under CP6 (MP6).
The client terminal is located next to CP3 (MP3).
This is a test scenario in which the client terminal is very close to CP3, in order to determine if the transmission with CP6, CP2 and CP1 is the same for the two devices. Please see the report files for more detail.
When against the walls, the terminal may not be in line of sight of the two CPs any more. Please see the report files for a precise indication of the terminal position.
At the first position, the client terminal is at 18 m from CP2 (18.20 m from the plastic door).
The client terminal is located at the intersection where CP1 sits (MP1). Please see the report files for the precise position.
The ore loading points are what appear as six short north-south-oriented tunnels in the southern tunnel where is CP6. In this scenario, the client terminal is in the second loading point from the east of the tunnel, that is between CP6 and CP3, at 13 m from CP6. It is in the middle (east-west) of the loading point, at 4.50 m north of the middle of the east-west tunnel (that is, about 5.50 m from the south wall). Obviously, the terminal is not in line of sight with either CP6 or CP3.
The client terminal is located in the recess at the south-west of CP3 (north wall of the south tunnel). More precisely, it is 12.20 m from CP3 and at about 0,75 m from the north and west walls of the recess. The west wall of the recess is at 13.50 m from CP3 (and therefore at 75 - 13.50 = 61.50 m from CP6). The terminal is in line of sight with CP3, but not with CP6 or CP2.
The client terminal is located in the section at the west of CP1 (MP11 and MP12). Please see the report files for the precise location.
This scenario is similar to the scenario 01, but the client terminal is carried by a human operator. Please see the report files for the description of the variations experimented.
In this scenario, the operator walks from the south-west end of the covered gallery (MP61, near the emergency exit), and goes all the way along the covered sections of tunnel to stop under CP5. In order to ease the treatment of the data, he stops next to each capture point and turns off the access point, hence preventing positioning requests to be sent. After about 30 seconds, the AP is started again, the operator waits for 5 to 6 seconds (the time needed by the client terminal to reconnect to the AP), then starts walking again.
See the report files for precise timing.
This scenario is similar to scenario 13, but the operator carries the mobile terminal from the middle of the main tunnel, facing the plastic door (at 6.65 m from CP3, 4.85 m from the door and 1.70 m from the west wall) to the water tank. The operator stops at about 0.50 m in front of the water tank.
The operator starts 40 m east of CP1, in line of sight and walks the tunnel west, in direction of CP1 and CP5, to end at the intersection at the west of CP5 (MP51).
The terminal is attached at 1.70 m above the ground level, and is at 0.30 m from the left side of the vehicle.
Since it is too big to go at the very end of the tunnel near the emergency exit, the vehicle starts from the western ore loading point (MP62). It then follows almost the same path as in scenario 22, stopping next to each capture point in the same way. After CP5, the vehicle continues to the west intersection (MP51).