Step 1: Configure Caddy web server

For DASH video streaming, we use the Caddy web server. It requires a configuration to serve the video segments from a file named Caddyfile.

The following Caddyfile is configured for 4-second video files in our network scenarios (refer to our publications) and is located at ~/mpquic-sbd-vagrant/Workspace/mpquic-sbd/Caddyfile.

1
2
3
4
5
6
7
8
9
10
11
12

# Unsecure listener for debugging purposes
http://10.0.2.2:4040 {
    root /home/vagrant/Workspace/video/4s
    log stdout
}

# Secure listener, required for TLS and QUIC connections
https://10.0.2.2:4242 {
    root /home/vagrant/Workspace/video/4s
    tls self_signed
    log stdout
}

Notes:

  • The path to the video files in line 3 and 9 should reflect the path inside the Vagrant VM, i.e., /home/vagrant/Workspace/mpquic-sbd/Caddyfile.
  • The Vagrant VM’s /home/vagrant/Workspace folder is mapped to your host at ~/mpquic-sbd-vagrant/Workspace/.
  • To run the Caddy web server with different video segments, just modify lines 3 and 9 to point to the correct subfolder, e.g., /home/vagrant/Workspace/video/2s for 2-second segments, or /home/vagrant/Workspace/video/10s for 10-second segments.

Step 2: Configure the Adaptive Birate (ABR) algorithms in the AStream client player

Navigate to the AStream implementation folder:”

1

cd ~/mpquic-sbd-vagrant/Workspace/mpquic-sbd/src/AStream/dist/client/

Next, in the config_dash.py file you can set the ABR parameters after line 52.

# Constants for the BASIC-2 adaptation scheme
BASIC_THRESHOLD = 1
BASIC_UPPER_THRESHOLD = 1.5
# Additional constant for the BASIC-3 adaptation scheme
BASIC_LOWER_THRESHOLD = 0.9
# Number of segments for moving average
BASIC_DELTA_COUNT = 20

# ---------------------------------------------------
# SARA (Segment Aware Rate Adaptation)
# ---------------------------------------------------
# Number of segments for moving weighted average
SARA_SAMPLE_COUNT = 5
# Constants for the Buffer in the Weighted adaptation scheme (in segments)
INITIAL_BUFFERING_COUNT = 1
RE_BUFFERING_COUNT = 1
ALPHA_BUFFER_COUNT = 1
BETA_BUFFER_COUNT = 2
# Set the size of the buffer in terms of segments. Set to unlimited if 0 or None
MAX_BUFFER_SIZE = 1

# ---------------------------------------------------
# Netflix (Buffer-based) ADAPTATION
# ---------------------------------------------------
# Constants for the Netflix Buffering scheme adaptation/netflix_buffer.py
# Constants is terms of buffer occupancy PERCENTAGE(%)
NETFLIX_RESERVOIR = 0.1
NETFLIX_CUSHION = 0.9
# Buffer Size in Number of segments 240/4
NETFLIX_BUFFER_SIZE = 60
NETFLIX_INITIAL_BUFFER = 2
NETFLIX_INITIAL_FACTOR = 0.875
# ...

Buffer size is determined by the number of segments based on the total video duration. Specifically:

  • The TBA (basic) ABR, by default, has a minimal buffer size (i.e., only 1 video segment), which results in a significant number of stall events during video playback.
  • The Hybrid (SARA) ABR uses the numbers of segments defined in ALPHA_BUFFER_COUNT and BETA_BUFFER_COUNT as thresholds to manage the buffer.
  • The BBA (Netflix) ABR buffer size is configured using the NETFLIX_BUFFER_SIZE parameter. In the example above, the buffer size is set to NETFLIX_BUFFER_SIZE = 60, assuming a total video duration of 240 minutes and a segment length of 4 seconds. Adjust NETFLIX_BUFFER_SIZE to reflect the total video duration in your experiments.

Updated: