We’ve all been there. You’re trying to watch a video on a public WiFi or at home where several people are watching different streams simultaneously. Then the videos start buffering, the pixilation comes in to compensate and it simply ruins the watching experience for everyone. Whether this is because of slow internet connection or simply a shoddy connection to the WiFi router, it is utterly demoralising. Now experts from MIT have created a tool to reduce video stream buffering to help multiple people share a limited WiFi connection.
The group from the Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed the Minerva system which analyses videos before playing them. This process then checks what the potential impact would be by playing that video at a lower quality.
Regular protocols on WiFi networks simply split the bandwidth available by the number of users connected to that router and line. So presumably, a phone connected to the WiFi being used to watch random YouTube videos is receiving the same bandwidth than someone trying to watch a live sports event, potentially in HD. As we’re aware, this will automatically need more bandwidth than a YouTube video or cartoon you’re showing to the kids.
According to the MIT experts, Minerva will then analyse both these videos in an offline phase (meaning they’re not being streamed yet) to see which could benefit from more bandwidth availability. More importantly, it will determine which could be served at a lower bandwidth without losing quality. It then assigns bandwidth accordingly. Could this be the optimal way to reduce video stream buffering with low bandwidth?
In real-world tests, Minerva was able to reduce rebuffering time almost by half. Furthermore, it was able to improve video playback quality in one third of cases to the equivalent shift from 720p to 1080p. While this is exciting, the more important implementation would be to do this across regions or internet services, not just a single WiFi hub.