TCP-Single-Conn-Perf

TCP Single Connection Performance

This folder contains the dataset, processing and plotting scripts for TCP single connection performance measurements conducted using 5GTracker and Azure VMs. It covers Figure 8 referred in Section 3.2 of the paper.

Folder Structure

Filename	Description
data/TCP-Single-Conn-Perf.csv	Contains description of the runs done
data/ping/az-[server_location].csv	Ping tests for each azure server
data/client/[iperf_run_number].json	Raw Iperf logs collected on client side
data/UE-Azure-Server-Distance.csv	CSV-based file containing distance between UE and Microsoft Azure servers tested
data-processed/TCP-Single-Conn-Perf_combined.csv	CSV-based combined file containing summarized results for all runs. See the Dataset Description section for more details
Process-Logs.py	Python script to process logs for TCP single connection performance
plot-section3-figure8.py	Python script to generate figure 8. See the Generating Plots section for more details

Dataset Description

Field Name	Description
server_location	Location of Microsoft Azure server in US
latency_min	Minimum latency seen for server in ping test (ms)
type	Connection type used (TCP-1 default, UDP, TCP-8, TCP-1 tuned)
iperf_run_number	Iperf run number in 5GTracker
distance	Distance between client and azure servers (miles)
throughput_rolled3_avg	Average throughput calculated after rolling with window size 3 (Mbps)
throughput_avg	Average throughput calculated from raw values (Mbps)
throughput_max	Maximum throughput seen for each run (Mbps)
throughput_90tile	90th percentile throughput seen for each run (Mbps)
throughput_95tile	95th percentile throughput seen for each run (Mbps)
throughput_median	Median throughput seen for run (Mbps)

Generating Plots

The scripts will generate Figure 8.

Requirements

Here are the software/package requirements. The version number in the bracket indicates the minimum version that our script has been tested on.

• Python 3 (3.7.7 and higher)
• Pandas (1.1.3 and higher)
• Matplotlib (3.3.1 and higher)

Running code

To regenerate the processed logs, the following command can be used.

python3 Process-Logs.py

The processed logs will be placed in data-processed folder. The Folder Structure section gives a detailed overview of all the files in data-processed folder.

To generate Figure 8 shown in the paper, simply run the following command

python3 plot-section3-figure8.py

This will create a plots folder having figures in 3 formats (png, pdf and eps).