Abstract

Dynamic Adaptive Streaming over HTTP (DASH) is becoming the de facto method for effective video traffic delivery at large scale. Its primer success factor returns to the full autonomy given to the streaming clients making them smarter and enabling decentralized logic of video quality decision at granular video chunks following a pull-based paradigm. However, the pure autonomy of the clients inherently results in an overall selfish environment where each client independently strives to improve its Quality of Experience (QoE). Consequently, the clients will hurt each other, including themselves, due to their limited scope of perception. This shortcoming could be addressed by employing a mechanism that has a global view, hence could efficiently manage the available resources. In this paper, we propose a game theoretical-based approach to address the issue of the client’s selfishness in multi-server setup, without affecting its autonomy. Particularly, we employ the coalitional game framework to affect the clients to the best server, ultimately to maximize the overall average quality of the clients while preventing re-buffering. We validate our solution through extensive experiments and showcase the effectiveness of the proposed solution.