Delay-Sensitive Rate Control Mechanisms for Personal Mobile Video Streaming Services개인 모바일 비디오 스트리밍 서비스를 위한 지연 민감형 레이트 제어 메커니즘

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 870
  • Download : 0
High-definition video streaming services are in growing demand in mobile network environments with the remarkable development of smartphones and wireless networks, such as LTE and WiFi. As these technologies evolve, various types of new wireless streaming services are attracting attention rather than down-link streaming services, such as mobile VoD. The representative service is a personal mobile streaming service in which the user’s mobile device is the center of real-time content provision. To provide such a service, a mobile device must handle encoding and streaming functions which are roles of a dedicated encoder and servers in a legacy streaming system. In addition, there is a technical challenge to deliver video data through not a broadcast-dedicated network but a best-effort wireless network, such as WiFi or LTE. In such a service, direct communication between two devices is performed for one-to-one streaming, such as drone racing control, requiring a strict real-time property. To stream to a large number of users, a streaming server that streams to a plurality of users must be provided. The bandwidths of current wireless networks vary. In many countries, broadband wireless access networks, such as 4G LTE, are not available nationwide, so handoffs are inevitable between networks with different characteristics depending on users’ mobility. This results in large changes in the network bandwidth and degrades the video streaming quality. In this dissertation, we propose two types of mobile streaming mechanisms. The first is an up-link streaming protocol for one-to-one mobile streaming service in which the mobile device is the center of providing real-time contents. To satisfy the real-time requirements, the size of the buffer must be minimized in the sender and receiver device, which causes serious deterioration in the video quality with even small changes in the network state. To solve this problem, we propose a NERV (network-aware encoding rate-control algorithm for real-time up-link video streaming services) protocol that predicts the state of the wireless network, adjusts the encoding rate in real-time according to the predicted network status, and streams it through the up-link interface. NERV predicts the network state without degrading the performance of the mobile device based on the inter-video frame delay value, which is the delay value between video frames measured at the receiving device. By comparing the predicted inter-video frame delay value with the three reference parameters (α, β and γ) defined by the algorithm, the encoding rate is adaptively adjusted and streamed through the up-link. We implemented NERV on a mobile device based on Google Android and measured performance on commercial network systems, such as LTE and WiFi, used by general users. Performance measurements in commercial LTE and WiFi networks as well as network emulation environments that artificially degrade network performance with predefined scenarios show that the average encoding rate increases by 35%, and the packet loss rate decreases by 45%. Second, we propose a MOSQUITO (mobile video streaming protocol for high-level QoE provisioning over heterogeneous wireless access networks), which is a buffer-based adaptive-bit-rate streaming protocol for down-link streaming service that transmits from server to multiple mobile devices for one-to-many streaming. MOSQUITO consists of a buffer-based adaptive-bit-rate (B2-ABR) algorithm that adaptively streams the network state according to the buffer size of the receiving device and a chunk file download-time-minimization (DTM) algorithm that minimizes the chunk file download time even network bandwidth is fluctuated severely. Through two algorithms, if the network state is good, the streaming bit rate is conservatively increased, and if the network condition is bad, the streaming rate is rapidly decreased. Also, in an environment in which only the minimum network bandwidth is provided higher than the lowest encoding rate of the chunk file, the chunk file can be downloaded within twice the target duration. This can reduce the screen stoppage due to buffer underflow when the wireless network bandwidth changes drastically. The protocol is transparently designed so that application developers who do not know how to operate the algorithm can easily use it. It is implemented in a commercial Google Android-based device for performance analysis and measurement in a commercial network system. Performance comparison with the MPEG-DASH protocol was performed in a heterogeneous wireless network environment in which 3G, 4G, and WiFi coexisted. Experiments demonstrated that the screen stoppage caused by buffer underflow is significantly reduced when a sudden change in bandwidth occurs due to handoff between different networks.
Advisors
Choi, Jun Kyunresearcher최준균researcher
Description
한국과학기술원 :정보통신공학과,
Publisher
한국과학기술원
Issue Date
2018
Identifier
325007
Language
eng
Description

학위논문(박사) - 한국과학기술원 : 정보통신공학과, 2018.2,[vii, 122 p. :]

URI
http://hdl.handle.net/10203/265379
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=734434&flag=dissertation
Appears in Collection
ICE-Theses_Ph.D.(박사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0