Short video has rapidly become one of the most popular online entertainment applications in the last two years. The new service pattern of short video network shows its unique characteristics, making it much more difficult to improve QoS by using traditional methods in content delivery network. In this paper, we deeply analyze the real traces of more than 28 million videos with 100 million accesses from 488 servers located in 33 cities, disclosing the following three unique characteristics/challenges of short video network compared to traditional Video on Demand (VoD) services: First, user access pattern shows less preferences on videos, making it's hard to find out the popular ones. Second, most videos have very few visits, which makes it difficult to learn and predict due to the lack of historical data. Third, the popularity changes much more quickly, which further makes it challenging to design efficient caching policy. To address these challenges, we're designing CoStore, a reinforcement learning-based caching policy which pre-fetches videos with the help of video correlations. This poster brings the short video network to light and we hope CoStore could also provide inspirations to related areas.
|Original language||English (US)|
|Title of host publication||INFOCOM 2019 - IEEE Conference on Computer Communications Workshops, INFOCOM WKSHPS 2019|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|Number of pages||2|
|State||Published - Apr 2019|
|Event||2019 INFOCOM IEEE Conference on Computer Communications Workshops, INFOCOM WKSHPS 2019 - Paris, France|
Duration: Apr 29 2019 → May 2 2019
|Name||INFOCOM 2019 - IEEE Conference on Computer Communications Workshops, INFOCOM WKSHPS 2019|
|Conference||2019 INFOCOM IEEE Conference on Computer Communications Workshops, INFOCOM WKSHPS 2019|
|Period||4/29/19 → 5/2/19|
Bibliographical noteFunding Information:
The work of Yuchao Zhang was supported in part by the China Postdoctoral Science Foundation under Grant 2018M630117, the National Natural Science Foundation of China under Grant 61802024, and the Huawei Autonomous and Service 2.0 Project under Grant A2018185.
- Caching Policy
- Content Delivery Network
- Short Videos