A Multi-Objective Video Crowdsourcing Method in Mobile Environment

• Main Authors: Chao Yan, Yuhao Chen, Fan Wang, Yiping Wen, Shucun Fu, Wanli Huang
• Format: Article
• Language: English
• Published: IEEE 2019-01-01
• Series: IEEE Access
• Subjects: Mobile video; D2D; crowdsourcing; DBSCAN
• Online Access: https://ieeexplore.ieee.org/document/8835905/

With the rapid development of mobile video services, HD and UHD videos are attractive for mobile users due to the realistic visual enjoyment and the accurate representation. However, the limited transmission bit rate in 4G communication network affects the experience of the users for watching videos. Crowdsourcing is considered as a reasonable and effective solution to alleviate the resource limitation. Through employing the crowdsourcing participants to download and transmit video segments, mobile users can get enhanced video services. However, it is still a significant challenge that how to avoid excessive payment and energy consumption when the crowdsourcing participants download the video segments for the mobile users. To address this challenge, a multi-objective video crowdsourcing method in mobile environment is proposed in this paper. Technically, the crowdsourcing participants apply device-to-device (D2D) communication technique rather than the cellular network or bluetooth transmission to transmit video segments to the mobile users. Here, we divide our problem into two situations, the single participant case and the multi-participants case. In the single participant case, we apply the improved dynamic programming algorithm to find strategies with more enhanced video service time that the crowdsourcing participants provide for the mobile users. Then Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) and Multiple Criteria Decision Making (MCDM) techniques are applied to find a balanced strategy to maximize the enhanced video service time and minimize the payment and the energy consumption. In the multi-participants case, through DBSCAN clustering, the problem with multi-participants is divided into several problems with single participant. Finally, extensive experimental evaluations are conducted to demonstrate the effectiveness and efficiency of our proposed method.