CRAWDAD cu/lte

LTE Measurements on a 100m (triangular) grid on the University of Colorado Campus.

This data was collected at the University of Colorado Boulder. It contains careful point measurements, taken on a 100m equilateral triangular lattice, of the Verizon LTE network.

date/time of measurement start: 2011-06-01

date/time of measurement end: 2011-09-01

collection environment: The environment is the campus of the University of Colorado, Boulder, Colorado, USA.

network configuration: The network was Verizon LTE network. All measurements were done "passively", without any greater access to the LTE network than a typical client device would have. 

data collection methodology: Data collection was performed using a netbook computer running MS Windows, JDSU (formerly Agilent) E6474A software, and an LG VL600 LTE USB dongle. At each point, a number of simultaneous channel performance metrics were collected, and a 60 second bidirectional throughput test was performed to an Internet host using iperf. The measurement devices were attached to a Croozer-brand bicycle/hand cart. At each point a GPS reading was taken from a hand-held Garmin GPS60. The data collection procedure is as follows:

1. Locate a point to measure using the map. Try to get as close to the actual point as possible. If it is on/in a building or other obstacle, go to the nearest accessible outdoor location and do the measurement there instead. If the point is totally unreachable (i.e. closest point is 100m away), then circle the point and move on.

2. Write down the ID of the point being measured (i.e., 12 4 0) for the 0th measurement at point 12 4 .

3. Using the hand held GPS device, write down the current location and accuracy of the GPS position.

4. Using the computer's clock (or a synchronized watch), write down the current time.

5. Click the green circle in the JDSU software to start a test.

6. After 5 seconds, a command window will open behind the main window and will run a throughput test. After 60 more seconds, this window will close.

7. Click the Stop button (or press Shift F9) to stop the test.

8. Mark off the location just measured on the map with an X.

9. Every third point, perform 2 additional ``cluster measurements'' where you select a nearby point within an 8 meter radius of the original point (20 wavelengths at 700 MHz). These points should be more or less selected at random.

At each point, the JDSU software collects a continuous stream of measurements using a large number of metrics. This data is stored in a proprietary file format that can be exported to a parseable ASCII CSV file after measurement is completed. The CSVs contain a number of metrics and a variable number of measurements per metric. These are documented to some extent in JDSU white-papers, but some are not documented. The header field should be sufficiently descriptive in most cases, however because the JDSU software and LG hardware are closed-source, it is not feasible to say how particular metrics are calculated.

limitation: The measurement device is a functional LTE radio and as such will connect to the best network available (ranking by RSRP) in order to pass traffic. A given measurement point may be served exclusively by a single cell, or by a combination of cells. Hence some number of measurements are available for most cells that serve the CU campus, but as a result of this behavior, the strongest cells receive the most measurements. And, the resulting measurement set is more descriptive of the combined coverage of the cells, rather than a complete model of any particular BS.

Traceset

cu/lte/jdsu

LTE Measurements on a 100m (triangular) grid on the University of Colorado Campus.

• file: cu/lte/crawdad.zip
• description: This data was collected at the University of Colorado Boulder. It contains careful point measurements, taken on a 100m equilateral triangular lattice, of the Verizon LTE network.
• methodology: Data collection was performed using a netbook computer running MS Windows, JDSU (formerly Agilent) E6474A software, and an LG VL600 LTE USB dongle. At each point, a number of simultaneous channel performance metrics were collected, and a 60 second bidirectional throughput test was performed to an Internet host using iperf. The measurement devices were attached to a Croozer-brand bicycle/hand cart. At each point a GPS reading was taken from a hand-held Garmin GPS60. The data collection procedure is as follows:

  1. Locate a point to measure using the map. Try to get as close to the actual point as possible. If it is on/in a building or other obstacle, go to the nearest accessible outdoor location and do the measurement there instead. If the point is totally unreachable (i.e. closest point is 100m away), then circle the point and move on.

  2. Write down the ID of the point being measured (i.e., 12 4 0) for the 0th measurement at point 12 4 .

  3. Using the hand held GPS device, write down the current location and accuracy of the GPS position.

  4. Using the computer's clock (or a synchronized watch), write down the current time.

  5. Click the green circle in the JDSU software to start a test.

  6. After 5 seconds, a command window will open behind the main window and will run a throughput test. After 60 more seconds, this window will close.

  7. Click the Stop button (or press Shift F9) to stop the test.

  8. Mark off the location just measured on the map with an X.

  9. Every third point, perform 2 additional ``cluster measurements'' where you select a nearby point within an 8 meter radius of the original point (20 wavelengths at 700 MHz). These points should be more or less selected at random.

  At each point, the JDSU software collects a continuous stream of measurements using a large number of metrics. This data is stored in a proprietary file format that can be exported to a parseable ASCII CSV file after measurement is completed. The CSVs contain a number of metrics and a variable number of measurements per metric. These are documented to some extent in JDSU white-papers, but some are not documented. The header field should be sufficiently descriptive in most cases, however because the JDSU software and LG hardware are closed-source, it is not feasible to say how particular metrics are calculated.

u/lte/jdsu Trace

• grid: LTE Measurements on a 100m (triangular) grid on the University of Colorado Campus. This data was collected at the University of Colorado Boulder. It contains careful point measurements, taken on a 100m equilateral triangular lattice, of the Verizon LTE network.

• configuration: Data collection was performed using a netbook computer running MS Windows, JDSU (formerly Agilent) E6474A software, and an LG VL600 LTE USB dongle. At each point, a number of simultaneous channel performance metrics were collected, and a 60 second bidirectional throughput test was performed to an Internet host using iperf. The measurement devices were attached to a Croozer-brand bicycle/hand cart. At each point a GPS reading was taken from a hand-held Garmin GPS60.
• format: The trace consists of a set of CSV files (one per measurement), locations of the measurements (locations.csv), and locations and information about the basestations (verizon.csv).