What are some of the tests we conducted

First Experiment  

During the initial stage of our experiment, we conducted latency tests on multiple internet providers to investigate how their network latency was affected under varying levels of usage - both high and low. 

Latency.mov

In this video, we demonstrate how latency can significantly impact the quality of video conferences, especially during activities that require real-time interaction like playing games. To illustrate this, my friend and I played the popular game "Rock Paper Seizures" while one of us intentionally used a network with high latency. As a result, you can see how even the slightest delay in data transfer can cause the video to lag, making it nearly impossible to play effectively.

Low Load Latency 

Heavy Load Latency 

Fido Internet

Virgin Plus Internet

The above figures display the results of our testing for various internet providers, showcasing the latency performance under both heavy and light network loads. By analyzing the data, we were able to observe how latency varies depending on the level of network usage.

Second Experiment 

In the second stage of our experiment, we conducted traffic tests between two Raspberry Pi devices to investigate the impact of bandwidth on network performance. By increasing the bandwidth during these tests, we were able to observe how the network handled larger amounts of traffic and whether latency issues arose.

For this experiment, we established a network configuration consisting of two Raspberry Pi devices connected to a TELUS gateway, which allowed us to run traffic between them and analyze the impact of different bandwidth levels on latency and network performance. Both Raspberry Pi devices were connected via ethernet cables, with the left device connected to the WAN and the right device connected via LAN.

10 Mb Bandwidth (Sender)

10 Mb Bandwidth (Receiver)

500 Mb Bandwidth (Sender)

500 Mb Bandwidth (Receiver)

1000 Mb Bandwidth (Sender)

1000 Mb Bandwidth (Receiver)

The accompanying graphs illustrate the significant impact of increasing bandwidth on network performance. For instance, when we limited the bandwidth to 10 Mb, the receiver received 10 Mb of data without any distractions or significant drops in bitrate. However, as we increased the bandwidth to 500 Mb, the network experienced some packet distractions and a noticeable drop in bitrate. These issues were further exacerbated when we increased the bandwidth to 1000 Mb, as depicted in the final graph. 

Third Experiment 

During the third stage of our experiment, we implemented Smart Quality Management to optimize network performance and reduce latency. By leveraging advanced traffic shaping techniques, we were able to ensure that network traffic was prioritized in a way that maximized bandwidth utilization and minimized delays.

A new device was added to set up to run SQM script and by using the Gigabyte Brix as a bridge, we were able to optimize network traffic between the Raspberry Pi devices, reducing latency and improving network performance. The addition of this device has significantly enhanced the capabilities of our network setup, enabling us to deliver a more reliable and responsive network connection.

In the following videos, we will observe how latency of each packet being transferred before and after implementing SQM. In the video there are two windows, the left one represents the command to run reverse traffic from the server to client, whereas the right window shows the time for each packet being sent to a server. Therefore when transferring reverse we will observe latency jump to high values (on the right window of the screen). 

IMG_7575.MOV

Sending one way traffic and observing the time it takes from server to client. 

IMG_7576.MOV

Sending eight data streams allows us to test the network performance with multiple streams of data, simulating real-life scenarios where multiple devices are sending and receiving data. 


The following graph represent the time it takes for a packets to be received by server (192.168.0.106)

Running 8 streams of data simulating real world scenario where multiple devices sending data.  

After Implementing Smart Queue Management (SQM)

IMG_7568.MOV
8tcp after

After implementing traffic shaping, there was a major reduction in the amount of latency in the network.

The implementation of traffic shaping has demonstrated significant reductions in delays, as evidenced. 

Last Experiment 

In the last stage we were able to add a TELUS Distribution Point Unit (DPU) was connected to a gateway to enable the transmission of fiber signals over non-fiber networks. To further optimize the network, an SQM script was written on OpenWRT, which is a Linux-based operating system running on the DPU.

The DPU was added into the network which resulted the current configuration of our test network. 

Before implementating the DPU, the bitrate was limited to 90 Mbit/s and the latency was around 1 ms. 

After implementating the DPU, the bitrate is limited to 300 Mbit/s and the latency is around 0.7 ms (for the higher bandwidths).