MIT develops the “Minerva” system that analyzes videos to adjust the required bandwidth and optimize viewing quality
Video now accounts for the majority of internet usage traffic in the world, and network equipment manufacturer Cisco predicts that it will reach 82% in 2022. The Massachusetts Institute of Technology (MIT) 's Computer Science and Artificial Intelligence Laboratory (CSAIL) optimizes congestion control using the player's state and file characteristics as a system suitable for supporting this massive video distribution. Minerva 'was developed. “Minerva” says it can significantly reduce buffering and pixelation without changing the underlying infrastructure.
A new tool allows multiple users in a house or network to watch video at the same time, with much less buffering and pixelation.— MIT CSAIL (@MIT_CSAIL) August 19, 2019
MIT's Minerva system (right), compared to a traditional congestion-control algorithm: https://t.co/IOpp2NLXbJ pic.twitter.com/RsbwPx5dKi
Better video streaming when WiFi is scarce | MIT CSAIL
MIT CSAIL's Minerva video protocol reduces buffering and pixelation
If you are using a video streaming service, there will be a difference in 'whether streaming is smooth' due to various factors such as differences in terminals, screen size, and files to be played. In fact, action movies and sports footage tend to require more bandwidth than lecture and speech footage.
However, for now, the mechanism of 'automatically increasing the allocated bandwidth because it is a sport' has not been made, and those who are watching sports videos can watch without stress, while those who are watching sports are clumsy A case of crying in reading occurs. Stanford University's Keith Winstein points out that this is 'fair from a technical point of view, but not connected to user experience or happiness'.
`` Minerva '' developed by MIT CSAIL has an analysis phase to determine how much the viewing quality will improve, for example, by allocating more bandwidth to sports coverage than speech video, It will improve the quality. This mechanism does not simply determine the bandwidth to be allocated for each genre, but also sees in which part of the movie the bandwidth is required, so even if the scene switches finely like a movie, appropriate communication The situation can be obtained and streaming can be comfortably viewed.
In the demonstration experiment, the playback quality improved from 720P to 1080P in 15% to 32% of the video, and the rebuffering that stopped the video halfway decreased by 47% on average.