Tamizhselvi S P, Muthuswamy Vijayalakshmi
Department of Computer Science & Engineering, Faculty of Information & Communication Engineering, KCG College of Technology, Anna University, Chennai, India.
Department of Information Science & Technology, Faculty of Information & Communication Engineering, College of Engineering Anna University, Chennai, India.
Peer Peer Netw Appl. 2021;14(4):2038-2060. doi: 10.1007/s12083-021-01134-1. Epub 2021 May 1.
In recent years, smartphone users are interested in large volumes to view live videos and sharing video resources over social media (e.g., Youtube, Netflix). The continuous streaming of video in mobile devices faces many challenges in network parameters namely bandwidth estimation, congestion window, throughput, delay, and transcoding is a challenging and time-consuming task. To perform these resource-intensive tasks via mobile is complicated, and hence, the cloud is integrated with smartphones to provide Mobile Cloud Computing (MCC). To resolve the issue, we propose a novel framework called delay aware bandwidth estimation and intelligent video transcoder in mobile cloud. In this paper, we introduced four techniques, namely, Markov Mobile Bandwidth Cloud Estimation (MMBCE), Cloud Dynamic Congestion Window (CDCW), Queue-based Video Processing for Cloud Server (QVPS), and Intelligent Video Transcoding for selecting Server (IVTS). To evaluate the performance of the proposed algorithm, we implemented a testbed using the two mobile configurations and the public cloud server Amazon Web Server (AWS). The study and results in a real environment demonstrate that our proposed framework can improve the QoS requirements and outperforms the existing algorithms. Firstly, MMBCE utilizes the well-known Markov Decision Process (MDP) model to estimate the best bandwidth of mobile using reward function. MMBCE improves the performance of 50% PDR compared with other algorithms. CDCW fits the congestion window and reduces packet loss dynamically. CDCW produces 40% more goodput with minimal PLR. Next, in QVPS, the M/M/S queueing model is processed to reduce the video processing delay and calculates the total service time. Finally, IVTS applies the M/G/N model and reduces 6% utilization of transcoding workload, by intelligently selecting the minimum workload of the transcoding server. The IVTS takes less time in slow and fast mode. The performance analysis and experimental evaluation show that the queueing model reduces the delay by 0.2 ms and the server's utilization by 20%. Hence, in this work, the cloud minimizes delay effectively to deliver a good quality of video streaming on mobile.
近年来，智能手机用户热衷于大量观看直播视频并通过社交媒体（如YouTube、Netflix）分享视频资源。移动设备中视频的持续流传输在网络参数方面面临诸多挑战，即带宽估计、拥塞窗口、吞吐量、延迟，并且转码是一项具有挑战性且耗时的任务。通过移动设备执行这些资源密集型任务很复杂，因此，云与智能手机集成以提供移动云计算（MCC）。为了解决这个问题，我们提出了一种名为移动云中延迟感知带宽估计和智能视频转码器的新颖框架。在本文中，我们介绍了四种技术，即马尔可夫移动带宽云估计（MMBCE）、云动态拥塞窗口（CDCW）、基于队列的云服务器视频处理（QVPS）以及用于选择服务器的智能视频转码（IVTS）。为了评估所提出算法的性能，我们使用两种移动配置和公共云服务器亚马逊网络服务（AWS）实现了一个测试平台。在实际环境中的研究和结果表明，我们提出的框架可以提高QoS要求并优于现有算法。首先，MMBCE利用著名的马尔可夫决策过程（MDP）模型通过奖励函数估计移动设备的最佳带宽。与其他算法相比，MMBCE将分组投递率（PDR）的性能提高了50%。CDCW拟合拥塞窗口并动态减少丢包。CDCW在最小分组丢失率（PLR）的情况下使吞吐量提高了40%。接下来，在QVPS中，处理M/M/S排队模型以减少视频处理延迟并计算总服务时间。最后，IVTS应用M/G/N模型，通过智能选择转码服务器的最小工作量，将转码工作量的利用率降低了6%。IVTS在慢速和快速模式下花费的时间更少。性能分析和实验评估表明，排队模型将延迟减少了0.2毫秒，服务器利用率降低了20%。因此，在这项工作中，云有效地最小化了延迟，以在移动设备上提供高质量的视频流。
