Fog Computing as an Enabler for Immersive Media: Service Scenarios and Research Opportunities

Abstract

Immersive media services, such as augmented reality and virtual reality (AR/VR), a 360degree video, and free-viewpoint video (FVV), are popular today. They require massive data storage, ultrahigh computing power, and ultralow latency. It is hard to fulfill these requirements simultaneously in a conventional communication system using a cloud/centralized radio access network (C-RAN). Specifically, due to centralized processing in such a system, the end-to-end latency, as well as the burden on the fronthaul network, are expected to be high. Fog computing-based radio access networks (F-RAN), in contrast, have been widely considered as an enabler for immersive media. Our contribution in this paper is threefold: First, we propose various service scenarios reflecting the characteristics of immersive media. Second, we identify the technologies that are required to support the proposed service scenarios under F-RAN and discuss how they can support the proposed scenarios efficiently. Third, we discuss possible research opportunities.a A list of acronyms can be found in the Appendix.INDEX TERMS Fog computing, radio access networks, immersive media, free-viewpoint video, 360-degree video, virtual reality, augmented reality.

Quality of Experience for Streaming Services: Measurements, Challenges and Insights

“…It is essential to keep in mind, that they can massively impact the brain, and affect its perceptions and reasoning, directly in an obvious manner (e.g., motion sickness, addiction, discomfort, eyestrain, nausea, migraine, etc.) [211]. Thus more studies have to consider these critical issues.…”

Artificial Intelligence Aided Joint Bit Rate Selection and Radio Resource Allocation for Adaptive Video Streaming over F-RANs

“…The first stage actor network has five layers with six inputs, including current channel quality, predicted channel quality, video chunk information, untransmitted chunk size, and buffer status. The number of neurons in each DNN layer is [6,128,64,32,1]. The hidden layers have a tanh activation function, while the output layer has a softmax function.…”

“…The hidden layers have a tanh activation function, while the output layer has a softmax function. The number of neurons for the second stage network in each layer is [7,256,128,64,32,1], where the output is the bit rate selection probability.…”

“…Forecasts indicate that by 2020, multimedia will require 90 percent of the global IP data traffic, where video delivery will account for 50 percent of the overall data stream via wireless communications. To satisfy the rapidly growing demand, the fog-computing-based radio access network (F-RAN) [1] has emerged as a key technology for delivering video media services, where storage and computation tasks can be offloaded from a centralized cloud center to fog nodes. Also, the fronthaul load is released and response time is reduced by caching content at the edge.…”

“…For the conventional cloud-based network, the controller of the ABR is deployed on the client side (video player), and only selects a proper video bit rate without the ability of adaptive resource allocation [9]. In contrast, an F-RAN provides an additional action dimension, namely radio resource allocation on the physical layer, for ABR streaming service, which is radio resource allocation on the physical layer [1]. Benefiting from the dedicated infrastructure, the fog-server (F-Server) is enabled to dynamically allocate radio resources, such as bandwidth, to improve the QoE of video service for each client.…”