Empirical Conditional Method: A New Approach to Predict Throughput in TCP Mobile Data Network
This work addresses the challenge of accurate and practical bandwidth prediction for video uploaders in mobile networks, offering incremental improvements over existing methods.
The paper tackled the problem of predicting future available bandwidth for live video streaming in TCP mobile data networks, proposing the Empirical Conditional Method (ECM) based on a Markov model, which reduced loss by about 10% compared to the Arithmetic Mean algorithm and improved bandwidth utilization without increasing loss or delay.
Experience of live video streaming can be improved if future available bandwidth can be predicted more accurately at the video uploader side. Thus follows a natural question which is how to make predictions both easily and precisely in an ever-changing network. Researchers have developed many prediction algorithms in the literature, from where a simple algorithm, Arithmetic Mean (AM), stands out. Based on that, we are purposing a new method called Empirical Conditional Method (ECM) based on a Markov model, hoping to utilize more information in the past data to get a more accurate prediction without loss of practicality. Through simulations, we found that our ECM algorithm performs better than the commonly used AM one, in the sense of reducing the loss by about 10% compared with AM. Besides, ECM also has a higher utilization rate of available bandwidth, which means ECM can send more data out while not having a higher loss rate or delay, especially under a low FPS setting. ECM can be more helpful for those who have relatively limited networks to reach a more considerable balance between frame loss rate and video quality hence improving the quality of experience.