Wireless Link Quality Estimation on FlockLab - and Beyond

This repository contains wireless link quality estimation data for the FlockLab testbed [1,2]. The rationale and description of this dataset is described in a the following abstract (pdf is included in this repository -- see below).

Dataset: Wireless Link Quality Estimationon FlockLab – and Beyond
Romain Jacob, Reto Da Forno, Roman Trüb, Andreas Biri, Lothar Thiele
DATA '19 Proceedings of the 2nd Workshop on Data Acquisition To Analysis, 2019

Data collection scenario

The data collection scenario is simple. Each FlockLab node is assigned one dedicated time slot. In this slot, a node sends 100 packets, called strobes. All strobes have the same payload size and use a given radio frequency channel and transmit power. All other nodes listen for the strobes and log packet reception events (i.e., success or failed). 

The test scenario is ran every two hours on two different platforms: the TelosB [3] and DPP-cc430 [4] platforms. We used all 27 nodes currently available.

Current dataset status

• 3 months of data with about 12 tests per day per platform
• 1 month of data with about 4 tests per day per platform

Data collection firmware

We are happy to share the link quality data we collected for the FlockLab testbed, but we also wanted to make it easier for others to collect similar datasets for other wireless networks. To achieve this, we include in this repository the data collection firmware we design. The entire data collection scheduling and control is done entirely in software, in order to make the firmware usable in a large variety on wireless networks. We implemented our data collection software using Baloo [5], a flexible network stack design framework based on Synchronous Transmission. Baloo efficiently handles network time synchronization and offers a flexible interface to schedule communication rounds. The firmware source code is available in the Baloo repository [6].

A set of experiment parameters can be patched directly in the firmware, which let the user tune the data collection without having to recompile the source code. This improves usability and facilitates automation. An example patching script is included in this repository. Currently, the following parameters can be patched:

• rf_channel,
• payload,
• host_id, and
• rand_seed

Current supported platforms

• TelosB [3]
• DPP-cc430 [4]

Repository versions

• v1.3.1
  Update abstract and notebook
• v1.3.0
  Addition of October 2019 data.
  The frequency of tests has been reduced to 4 per day, executing at (approximately) 1:00, 7:00, 13:00, and 19:00.
  From October 28 onward, time shifted by one hour (2:00, 8:00, 14:00, 20:00).
• v1.2.0
  Addition of September 2019 data.
  Many missing tests on the 12, 13, 19, and 20 of September (due to construction works in the building).
• v1.1.4
  Update of the abstract to have hyperlinks to the plots. Corrected typos.

• v1.1.0
  Initial version.
  Add the data collected in August 2019.
  Data collected was disturbed at the beginning of the month and resumed normally on the August 13. Data from previous days are incomplete.
• v1.0.0
  Initial version.
  Contain collected data in July 2019, from the 10th to 30th of July.
  No data were collected on the 31st of July (technical issue).

List of files

• yyyy-mm_raw_platform.zip
  Archive containing all FlockLab test result files (one .zip file per month and per platform).
• yyyy-mm_preprocessed_all.zip
  Archive containing preprocessed csv files, one per month and per platform.
• firmware.zip
  Archive containing the firmware for all supported platform.
• firmware_patch.sh
  Example bash script illustrating the firmware patching.
• parse_flocklab_results.ipynb [open in nbviewer]
  Jupyter notebook used to create the pre-process data files. Also includes some example of data visualization.
• parse_flocklab_results.html
  HTML rendering of the notebook (static).
• plots.zip
  Archive containing high resolution visualization of the dataset, generated by the parse_flocklab_results notebook, and presented in the abstract.
• abstract.pdf
  A 3 page abstract presenting the dataset.
• CRediT.pdf
  The list of contributions from the authors.

Dataset updates

We plan to carry-on collecting this data in the long-run (However, we will likely decrease the time resolution to free testing time up). The newly collected data will be added to this;repository periodically; we envision monthly updates.

References

[1] R. Lim, F. Ferrari, M. Zimmerling, C. Walser, P. Sommer, and J. Beutel, “FlockLab: A Testbed for Distributed, Synchronized Tracing and Profiling of Wireless Embedded Systems,” in Proceedings of the 12th International Conference on Information Processing in Sensor Networks, New York, NY, USA, 2013, pp. 153–166.

[2] “FlockLab,” GitLab. [Online]. Available: https://gitlab.ethz.ch/tec/public/flocklab/wikis/home. [Accessed: 24-Jul-2019].

[3] Advanticsys, “MTM-CM5000-MSP 802.15.4 TelosB mote Module.” [Online]. Available: https://www.advanticsys.com/shop/mtmcm5000msp-p-14.html. [Accessed: 21-Sep-2018].

[4] Texas Instruments, “CC430F6137 16-Bit Ultra-Low-Power MCU.” [Online]. Available: http://www.ti.com/product/CC430F6137. [Accessed: 21-Sep-2018].

[5] R. Jacob, J. Bächli, R. Da Forno, and L. Thiele, “Synchronous Transmissions Made Easy: Design Your Network Stack with Baloo,” in Proceedings of the 2019 International Conference on Embedded Wireless Systems and Networks, 2019.

[6] “Baloo,” Dec-2018. [Online]. Available: http://www.romainjacob.net/research/baloo/.