标题: 行動廣播網路之資料復原研究
The Data Recovery Issue in Mobile Broadcast Networks
作者: 楊文新
Yang, Wen-Hsin
曾煜棋
林寶樹
Tseng, Yu-Chee
Lin, Bao-Shuh
資訊科學與工程研究所
关键字: 廣播網路;資料復原;DVB-H;DVB-IPDC;能源效率;換手;網路編碼;網路選擇;復原網路;WiFi;WiMAX;無線網路;broadcast;data recovery;DVB-H;DVB-IPDC;energy efficiency;handover;network coding;network selection;recovery network;WiFi;WiMAX;wireless network
公开日期: 2011
摘要: DVB-H (digital video broadcasting-handheld)及DVB-IPDC (IP datacast over DVB-H)近年來已發展成數位行動廣播服務的規格。其中DVB-H提供為手持行動裝置上進行數位視訊廣播,而DVB-IPDC則可結合另外的無線網路(如IP中繼網路),進行DVB-H廣播資料遺失補償處理。本文以採用無線寬頻網路WiMAX及WiFi之DVB-IPDC架構基礎,由滿足行動使用者服務品質的需求觀點,研究行動廣播網路之資料復原議題。研究項目分為:一、討論透過復原網路進行行動廣播資料復原的方法;二、討論行動裝置在復原網路內移動產生換手的處理方法。以下分別說明。

首先透過復原網路進行行動廣播資料復原方法的研究,討論DVB-H廣播內容資料遺失或錯誤時,由一個IP中繼網路協助進行有效率的資料重送處理。提出GPL (group packet loss)及BDH (broadcast data handover)兩個問題。GPL問題發生於重送相同的廣播資料需求,由大量的行動裝置提出,而這些行動裝置散佈於鄰近位置呈現地域性關係,或者重送廣播資料具有時間連續性的關係。BDH問題則發生於行動裝置在之前所在的服務網路細胞範圍內提出重送廣播資料需求後,因為移動性換手到新的服務網路細胞範圍,而重覆提出重送廣播資料需求。為了解決GPL及BDH問題,本文提出透過發掘時間與空間相互關連性之lazy wait及group acknowledgement方法與機制,以減少重覆提出重送廣播資料的需求數量,也可以避免行動裝置因為換手重覆提出重送廣播資料需求。此外,系統根據行動裝置上的DVB-H接收訊號品質進行數學分析,動態調整系統參數lazy wait計時器(timer)及group acknowledge回應判斷條件臨界值。模擬結果顯示所提方法可以達到減少重送廣播資料需求及重送廣播資料數量的效果,並可減輕造成IP中繼網路擁擠的狀況。

其次討論DVB-H與WiMAX行動廣播網路整合架構下,利用網路編碼進行廣播資料隨需(on-demand)復原方法的研究。此時行動裝置可以保持接收DVB-H廣播視訊內容,同時使用WiMAX網路做為復原管道進行重送廣播資料。此復原管道建模成一個提送時段接著一個復原時段的重覆訊框,行動裝置在提送時段內提出重送廣播資料需求到行動廣播伺服器,在復原時段內需重送的錯誤或遺失廣播資料則透過WiMAX網路傳遞到行動裝置。本文提出XOR編碼基礎的優先權網路編碼方法以達到復原更多具時效性廣播資料的目標,包括RGS (recovery by greedy selection)及RPS (recovery by prioritized selection)兩種方法。RGS透過建構權重二分圖來表達編碼資料封包(coded packet)對資料復原的影響程度,採用貪婪策略找尋最大權重最小覆蓋組合(maximum weight minimum dominating set)的編碼資料封包組。RPS延伸RGS方法,考量重送錯誤或遺失廣播資料與提出之重送廣播資料需求兩者之間數量的比重關係,以及編碼資料封包的重送順序兩項因素,以樹狀回溯模式進行權重調整並找尋最大權重最小覆蓋組合的編碼資料封包組。模擬結果顯示所提方法可以提升錯誤或遺失廣播資料之復原效率,並可減少需重送廣播資料因為逾越有效時限而無法復原的數量。

最後行動裝置在WiMAX與WiFi整合的復原網路內移動產生換手問題的研究。此復原網路中之WiFi網路雖然已經廣佈於大樓及公共空間區域提供無線上網,但其信號覆蓋區域有限,無法提供行動裝置於移動中之無縫連結要求。因此透過具有較大信號覆蓋區域的WiMAX網路來協助行動裝置進行換手處理,其中節省電能為一重要議題。本文根據行動裝置過去的換手紀錄,提出HGMA (handover scheme with geographic mobility awareness)方法,透過三個方式達到節省電能目標。一、透過度量行動裝置對網路的接收訊號強度及其平均移動速度,避免行動裝置觸發不必要的換手需求。二、換手時利用HCS (handover candidate selection)篩選候選網路組,以減少掃描網路接取點而節省電能。三、增加使用原先連結的WiFi或WiMAX網路類型接取點機會,減少因為使用不同類型網路切換通訊介面的頻率以節省電能。HGMA也可以讓行動裝置換手時較偏好於使用WiFi網路及獲得頻寬保證。模擬結果為行動裝置執行換手時可以節省59~80%電量消耗,獲得16~62%的更多機會使用WiFi網路,同時增加20~61%的機率滿足應用頻寬需求的服務品質要求。

總結,本文以結合無線寬頻網路的DVB-IPDC架構,研究行動廣播內容資料遺失或錯誤的資料復原處理方法。透過復原網路進行有效率的廣播資料重送,同時行動裝置在WiMAX與WiFi整合的復原網路間換手時耗損較低的電能。
Recently, DVB-H (digital video broadcasting{handheld) and DVB-IPDC (IP datacast over DVB-H) have been developed to support mobile broadcasting services. DVB-H is designed to support digital video broadcast for handheld devices, while DVB-IPDC tries to cooperate with an wireless network such as IP-relay network to complement the data loss problem in DVB-H. Assuming that some wireless broadband networks such as WiMAX and WiFi networks are adopted to support DVB-IPDC, this dissertation will study the data recovery issue in the mobile broadcast network from the perspective of satisfying quality of service to mobile users. In particular, we break down two genres of tasks: one genre of tasks is to conduct the data recovery through a recovery network, the other genre of tasks is to discuss recovery network selection for the mobile devices.

For conducting the data recovery through a recovery network, we first discuss the message-efficient data recovery for DVB-H data losses through an IP-relay network, which points out two critical problems: group packet loss (GPL) and broadcast data handover (BDH). GPL occurs when there is a burst of retransmission requests for the same pieces
of data with high spatial or temporal correlation. BDH happens when some devices that made the above requests handover to new serving cells. To solve these problems, we propose a lazy wait and a group acknowledgement schemes to alleviate duplicate requests by exploiting their spatial and temporal correlation. This not only reduces the requests submitted by neighboring devices in both space and time domains, but also avoids handovering devices from sending duplicate requests in new cells. Through mathematical analysis, we show how to adaptively adjust the timers of lazy wait and group acknowledgement based on channel quality. Simulation results prove that our schemes
can e±ciently reduce retransmission requests and retransmission packets, thus alleviating congestion in the IP-relay network.

Then, we consider the on-demand data recovery by network coding for mobile broadcasting systems, in which studies an integrated DVB-H and WiMAX mobile broadcasting architecture. DVB-H continually broadcasts videos to mobile devices. WiMAX serves as a recovery channel for mobile devices to request for retransmissions, which is modeled by
repetitive frames, each with a submission period followed by a recovery period. Mobile devices submit their requests in the submission periods while the WiMAX network sends the lost packets in the recovery periods. To recover more urgent packets first, we develop
two prioritized network coding schemes based on XOR coding. The first recovery by greedy selection (RGS) scheme constructs a weighted bipartite graph to reflect the influence of each coded packet and then adopts a greedy strategy to ‾nd a minimum dominating set of coded packets. By extending the similar concept to cope with the relation between requests submissions and packets losses as well as the retransmission sequence of coded packets, the second recovery by prioritized selection (RPS) scheme adopts a tree traversal approach to adjust weights and ‾nd a minimal set of coded packets from the graph. Simulation results verify that both RGS and RPS can improve recovery e±ciency while reduce the number of packets being dropped due to missing deadlines.

For the recovery network selection for the mobile devices, we discuss the energy-efficient concern in an integrated WiMAX and WiFi network for the DVB-H mobile devices. To provide wireless Internet access, WiFi networks have been deployed in many regions such as buildings and campuses. However, WiFi networks are still insufficient to support ubiquitous wireless service due to their narrow coverage. One possibility to resolve this de‾ciency is to integrate WiFi networks with the wide-range WiMAX networks. Under such an integrated WiMAX and WiFi network, how to conduct energy-efficient handovers is a critical issue. In this dissertation, we propose a handover scheme with geographic mobility awareness (HGMA), which considers the historical handover patterns of mobile devices. HGMA can conserve the energy of handovering devices from three aspects. Firstly, it prevents mobile devices from triggering unnecessary handovers accord-
ing to their received signal strength and moving speeds. Secondly, it contains a handover candidate selection (HCS) method for mobile devices to intelligently select a subset of WiFi access points or WiMAX relay stations to be scanned. Therefore, mobile devices can reduce their network scanning and thus save their energy. Thirdly, HGMA prefers mobile devices staying in their original WiMAX or WiFi networks. This can prevent mobile devices from consuming too much energy on interface switching. In addition, HGMA prefers the low-tier WiFi network over the WiMAX network and guarantees the bandwidth requirements of handovering devices. Simulation results show that HGMA can save about 59% to 80% of energy consumption of a handover operation, make mobile devices to associate with WiFi networks with 16% to 62% more probabilities, and increase about 20% to 61% of QoS satisfaction ratio to handovering devices.

In short, we pursue data recovery schemes to the mobile devices for DVB-H data losses by partnering with wireless broadband networks, which are able to do efficient data retransmissions for error recovery via a recovery network and achieve less energy consumption on the mobile devices for handovering between WiMAX and WiFi networks.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079317803
http://hdl.handle.net/11536/40547
显示于类别:Thesis


文件中的档案:

  1. 780301.pdf

If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.