The internet-based telephony has dramatic risen in recent years, especially for long-distance international calling. However, the real-time interactive nature of audio calls make them far more sensitive to issue such as RTT (round-trip time), packet loss as well as jitter induced by network.
Therefore, a main research problem has been concentrated on by this paper is:
how to alleviate the long-distance internet-based calling’s quality problem and improve the user-perceived call quality.
- The auxiliary research problem:
In order to analyze and study the above main research problem (how to improve the user-perceived call quality), the author need to firstly analyze and study two questions:
(1) how network performance impacts user-perceived voice call quality, and (2) why and where such quality problems occur in the wild.
- The proposal architecture to alleviate calling’s quality problem:
This paper presents a VIA architecture that revisits the use of classical overlay techniques to relay calls.
- The challenge of VIA be realized actually in practice:
be in the face of significant spatial and temporal variability in performance and the existence of a large number of relaying choices.
- Solution to the above challenge:
This paper proposes a practical relay selection approach that intelligently combines prediction-based filtering with an online exploration- exploitation strategy, which refer to as prediction-guided exploration approach.
- The benefits of prediction-guided exploration:
Prediction-guided exploration, thus, strikes a balance between exploration-based approaches which explore all possible choices, and prediction-based approaches which attempt to predict a single best choice from history information.
- Approach of solving the above auxiliary research problem:
through analyzing a data set of 430 million calls from Skype, with clients spread across 1900 ASes and 126 countries.
- The writing’s logic in this paper:
Due to the real-time interactive nature of audio calls, internet-based callings are far more sensitive to issues such as RTT, packet loss, as well as jitter induced by network.
1)Firstly, in order to analyze the impact of the above network performances on user-perceived calling’s quality, the author analyzes and study a data set of 430 million calls from Skype. The key observations from this section are:
- Network performance matters. User experience of calls is impacted by even small changes in network metrics.
- Wide-area communication, such as international and inter-domain calls, are more prone to bad network performance, and have a large room of improvement.
- Calls suffering from poor networks are spread spatially and temporally. Most calls with poor network performance are not from a handful of source-destination AS pairs. And most source-destination pairs only experience high PNR for a relatively short period of time.
2) Secondly, to improve internet-based calling’s quality affected by the above poor network performances, the author proposes an architecture called VIA that revisits the use of classical overlay techniques to relay calls.
Besides, in order to make readers learn about benefits of this architecture intuitively, the author quantifies the potential benefits of a managed overlay network for improving audio call quality. Trace-driven analysis shows that VIA can potentially improve up to 53% of calls whose quality is impacted by poor performance on key network metrics.
3) Highlighting the challenges in achieving these benefits and presenting a practical relay selection algorithm that delivers close-to-optimal performance.