Wireless Network Traffic Time Series of an Enterprise Network

Citation Author(s):
Su Pyae
Sone
University of Oulu, CWC-RT
Janne
Lehtomäki
University of Oulu, CWC-RT
Zaheer
Khan
University of Oulu, CWC-RT
Submitted by:
Su Pyae Sone
Last updated:
Wed, 05/06/2020 - 11:44
DOI:
10.21227/wnsx-ja20
Data Format:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

This MATLAB dataset (.mat) contains the collected real measurement data from a total of 470 access points (APs) deployed in the Linnanmaa campus of the University of Oulu, Finland. The measurements include IDs, dates of data collection, number of users, received traffic data, transmitted traffic data and location names of each AP. Each observation of traffic data and number of users provide the data value at every 10-minute interval between December 18, 2018 and February 12, 2019. Please cite this as: S. P. Sone & Janne Lehtomäki & Zaheer Khan. "Wireless Traffic Usage Forecasting Using Real Enterprise Network Data: Analysis and Methods". supplementary material, Dataset, 2020. Size: 25.1 MB (26,338,384 bytes). Platform and Environment: any platform supporting MATLAB .mat files (such as MATLAB, Octave, etc.) 

Instructions: 

Major component description: There are 3 main major components: number of users connected at collected time (numb_users), received traffic data in bytes (rxbytes) and transmitted traffic data in bytes (txbytes) of each AP in this dataset. Dates and times of data collection (date) can be converted into the serial date number by using datenum() function in MATLAB.

 

Received and transmitted traffic data are in the cumulative time series format so that differencing every 2 consecutive observations is required to get the actual values at every 10-minute. It can be done by using diff() function in MATLAB, for example, "diff(ap184016.txbytes)".

 

Setup and running instructions: First, MATLAB must be installed in the computer correctly. Then, the downloaded dataset should be placed in the folder whose path is already specified in MATLAB (see https://in.mathworks.com/help/matlab/matlab_env/specify-file-names.html).

 

Once the dataset (APs_dataset.mat) is loaded correctly in MATLAB, total 470 structure arrays with the IDs of each AP will appear in MATLAB Workspace. Then, the desired time series can be called in MATLAB, for example, "Tx_data = diff(ap184016.txbytes);".

Dataset Files

ACCESS ON AWS

Documentation

AttachmentSize
PDF icon README.pdf164.8 KB