Awesome

A LEAF User Interface with Streamlit

This project is for COMPSCI5018P MSc Development Project for IT+ at the University of Glasgow. Given the limit of the development cycle and the unfamiliarity of the framework used by the developer, problems are found from the industrial application point of view.

LEAF-UI is based on practising the introductory examples in the original LEAF example in the python version, users can directly watch the power usage in user-modified infrastructure and application chart. This application is conducted through streamlit for fast-built data representation.

Some new methods were introduced in the leaf-GUI for agile and compatibility usage; as a result, users will no longer need to write complex loops for writing to csv, and modify how each data would be displayed in the final chart.

🌐 Rapid Deployment

The permanent link of this application is https://zzzzzzzzzed-leaf-gui-leaf-akha61.streamlitapp.com/.

LEAF-UI is hosted on Streamlit Cloud, if you touch LEAF-UI through this link, please copy the content in cloud_setter.py into the Python compiler on the cloud for the first time to prevent errors. The reasons have been discussed in detail in the project report.

⚙️ Get started on your local machine!

Set up your environment and run:

streamlit run LEAF.py

🚀 How does it work?

For LEAF starters, you can just input examples in the python interpreter on the home page. Then, click 'run simulator' and check the results.

If you are running your unique script:

Please do not forget to add this line for you to easily draw and classify lines in the graph for each power meter.

import csv_handler as ch

To understand why we are importing csv_handler, use the original smart_city_traffic code snippet as an example：

Set power meters:

env.process(pm_cloud.run(env))
env.process(pm_fog.run(env))
env.process(pm_wan_up.run(env))
env.process(pm_wan_down.run(env))
env.process(pm_wifi.run(env))

Then

Original method write results to csv:

csv_content = "time,cloud static,cloud dynamic,fog static,fog dynamic,wifi static,wifi dynamic,wanUp static," \
                      "wanUp dynamic,wanDown static,wanDown dynamic\n"
for i, (cloud, fog, wifi, wan_up, wan_down) in enumerate(zip(pm_cloud.measurements, pm_fog.measurements, pm_wifi.measurements, pm_wan_up.measurements, pm_wan_down.measurements)):
csv_content += f"{i},{cloud.static},{cloud.dynamic},{fog.static},{fog.dynamic},{wifi.static},{wifi.dynamic},{wan_up.static},{wan_up.dynamic},{wan_down.static},{wan_down.dynamic}\n"
with open(f"{result_dir}/infrastructure.csv", 'w') as csvfile:
            csvfile.write(csv_content)

The above code block shows the efforts made in the original script to visualize some power meters; however, this code is not reusable and complex. Moreover, even with continued use of this method, some LEAF simulator characteristics such as sampling frequency and delay are difficult to solve.

So, a more convinent method we suggest here would be:

ch.output_csv(PM=pm_cloud, rename='Cloud',type = 1)
ch.output_csv(PM=pm_fog, rename='Fog',type = 1)
ch.output_csv(PM=pm_wan_up, rename='Wan up',type = 1)
ch.output_csv(PM=pm_wan_down, rename='Wan down',type = 1)
ch.output_csv(PM=pm_wifi, rename='WIFI',type = 1)

The output_csv method takes three parameters:

(PM = power meter instance, rename = 'name you want in the graph', type = 1 is for infrastructure type = 2 is for application)

Finally, don't forget to add

ch.merge_results()

⚠️ Risks

Because of the existing defects of streamlit, if there is a cell with the value NAN in a column, the whole column will disappear on the graph. Therefore, this project fills NAH with valid upward values to compensate for the missing data caused by different sampling intervals. This may lead to a bias between the final result and the actual power usage in practical applications. However, according to the statistics principle: the larger the sample, the smaller the error. We can conclude that when the user reduces the sampling frequency, the user also tolerates and accepts data error since users decide the value of the interval. So there is not much of a real problem with our approach that needs to be discussed.

For better conducting the living chart, it is best not to simulate more than 3600 seconds, which is a simulated hour, because a large amount of data flow may cause the page to get stuck, but after waiting for a period of calculation time, the final result will be displayed correctly. Suppose you must simulate the power consumption in an extensive time background, such as 3600*24, to find out the difference between day and night. In that case, you can get the CSV file through original LEAF and import the CSV file on the home page by the ‘Import Results’ button, and the development will be displayed correctly.

❓ More information

To find out what is LEAF, what is LEAF used for, please click LEAF.