基于云存儲的視頻信息分布式優(yōu)化處理系統(tǒng)的研究與設計
發(fā)布時間：2019-06-05 12:21
【摘要】：隨著科學技術的進步,視頻處理系統(tǒng)雖然得到一定發(fā)展,但面對需要滿足大量訪問、快速響應等高質量視頻服務時,傳統(tǒng)方案在體系結構和負載均衡等方面還不夠成熟,已經(jīng)不能再滿足當今的需求。本文所討論的基于云存儲的視頻信息分布式優(yōu)化處理系統(tǒng),其中云存儲作為一個新興的研究和應用領域,其具有快速部署,低成本,靈活調(diào)整規(guī)模等優(yōu)勢,但云存儲同樣也受到了一定的限制,原因在于我們雖然擁有一系列對負載均衡進行衡量的算法,但是由于不能提前對負載進行預算度量,這就使負載均衡失去了基礎,限制了整個系統(tǒng)性能。基于小波神經(jīng)網(wǎng)絡的負載均衡具有可預測性和自學習性,使負載均衡達到合理應用性。 本文就如何構建云存儲環(huán)境、如何優(yōu)化視頻信息處理技術和如何運用小波神經(jīng)網(wǎng)絡來處理負載均衡策略這三個方面,給出了基于云存儲的視頻信息分布式優(yōu)化處理系統(tǒng)的設計方案,在某種程度上解決傳統(tǒng)視頻信息處理系統(tǒng)技術上的不足,大大簡化其應用環(huán)節(jié),實現(xiàn)視頻信息資源充分共享,提高其利用效率。本課題的主要研究工作如下： (1)云存儲構架的研究與設計。基于云存儲概念及特點,設計了云存儲四層存儲服務器模型,從底層到上層依次是：云存儲層,數(shù)據(jù)管理層,應用接口層(也叫數(shù)據(jù)服務層)以及用戶訪問層。本文提供的設計方案為：利用普通PC機群搭建云存儲中的底層-云存儲層,采用多種功能模塊分塊管理進行數(shù)據(jù)管理層的設計,在應用接口層針對相應功能開發(fā)一些實際接口,方便與訪問層用戶操作的交互。 (2)視頻信息分布式優(yōu)化處理�；谠拼鎯Νh(huán)境,系統(tǒng)分別從視頻信息傳輸、調(diào)度、存儲等方面進行優(yōu)化設計。對接收到的視頻信息進行重組及H.264解碼,采用TCP與RTP相結合的方式進行傳輸。在調(diào)度方面,選擇一種新調(diào)度算法-最強能力優(yōu)先調(diào)度算法,存儲策略則是采用基于時間序列的視頻文件熱度進行有效存儲。 (3)負載均衡的研究與設計。針對傳統(tǒng)算法的局限性,文章提出了一種基于小波神經(jīng)網(wǎng)絡預測模型的改進算法,并在MATLAB環(huán)境下進行仿真實驗,證明優(yōu)越性。并以此為基礎,設計了系統(tǒng)負載均衡策略。
[Abstract]:With the progress of science and technology, although video processing system has been developed to a certain extent, the traditional scheme is not mature enough in architecture and load balancing when it needs to meet a large number of access, fast response and other high-quality video services. It can no longer meet the needs of today. The distributed optimal processing system of video information based on cloud storage discussed in this paper, in which cloud storage, as a new research and application field, has the advantages of rapid deployment, low cost, flexible adjustment of scale and so on. But cloud storage is also limited because although we have a series of algorithms to measure load balancing, because we can not measure the load ahead of time, load balancing has lost its foundation. The performance of the whole system is limited. Load balancing based on wavelet neural network has predictability and self-study habit, which makes load balancing reach reasonable application. This paper focuses on how to construct cloud storage environment, how to optimize video information processing technology and how to use wavelet neural network to deal with load balancing strategy. The design scheme of video information distributed optimization processing system based on cloud storage is given, which solves the technical shortcomings of traditional video information processing system to some extent, greatly simplifies its application links, and realizes the full sharing of video information resources. Improve its utilization efficiency. The main research work of this paper is as follows: (1) the research and design of cloud storage architecture. Based on the concept and characteristics of cloud storage, a four-layer storage server model of cloud storage is designed. From the bottom layer to the upper layer, the cloud storage layer, the data management layer, the application interface layer (also known as the data service layer) and the user access layer are designed. The design scheme provided in this paper is as follows: using ordinary PC cluster to build the bottom layer of cloud storage, using a variety of functional modules to manage the data management, and developing some practical interfaces for the corresponding functions in the application interface layer. Facilitate interaction with access layer user operations. (2) distributed optimal processing of video information. Based on cloud storage environment, the system is optimized from the aspects of video information transmission, scheduling, storage and so on. The received video information is reorganized and H.264 decoded, and TCP and RTP are combined to transmit the video information. In the aspect of scheduling, a new scheduling algorithm, the strongest capability priority scheduling algorithm, is selected, and the storage strategy is to store the video file heat based on time series effectively. (3) Research and design of load balancing. Aiming at the limitation of traditional algorithm, this paper proposes an improved algorithm based on wavelet neural network prediction model, and carries on the simulation experiment in MATLAB environment to prove the superiority. On this basis, the load balancing strategy of the system is designed.
【學位授予單位】：廣東工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2012
【分類號】：TP391.41;TP333
本文編號：2493525
[Abstract]:With the progress of science and technology, although video processing system has been developed to a certain extent, the traditional scheme is not mature enough in architecture and load balancing when it needs to meet a large number of access, fast response and other high-quality video services. It can no longer meet the needs of today. The distributed optimal processing system of video information based on cloud storage discussed in this paper, in which cloud storage, as a new research and application field, has the advantages of rapid deployment, low cost, flexible adjustment of scale and so on. But cloud storage is also limited because although we have a series of algorithms to measure load balancing, because we can not measure the load ahead of time, load balancing has lost its foundation. The performance of the whole system is limited. Load balancing based on wavelet neural network has predictability and self-study habit, which makes load balancing reach reasonable application. This paper focuses on how to construct cloud storage environment, how to optimize video information processing technology and how to use wavelet neural network to deal with load balancing strategy. The design scheme of video information distributed optimization processing system based on cloud storage is given, which solves the technical shortcomings of traditional video information processing system to some extent, greatly simplifies its application links, and realizes the full sharing of video information resources. Improve its utilization efficiency. The main research work of this paper is as follows: (1) the research and design of cloud storage architecture. Based on the concept and characteristics of cloud storage, a four-layer storage server model of cloud storage is designed. From the bottom layer to the upper layer, the cloud storage layer, the data management layer, the application interface layer (also known as the data service layer) and the user access layer are designed. The design scheme provided in this paper is as follows: using ordinary PC cluster to build the bottom layer of cloud storage, using a variety of functional modules to manage the data management, and developing some practical interfaces for the corresponding functions in the application interface layer. Facilitate interaction with access layer user operations. (2) distributed optimal processing of video information. Based on cloud storage environment, the system is optimized from the aspects of video information transmission, scheduling, storage and so on. The received video information is reorganized and H.264 decoded, and TCP and RTP are combined to transmit the video information. In the aspect of scheduling, a new scheduling algorithm, the strongest capability priority scheduling algorithm, is selected, and the storage strategy is to store the video file heat based on time series effectively. (3) Research and design of load balancing. Aiming at the limitation of traditional algorithm, this paper proposes an improved algorithm based on wavelet neural network prediction model, and carries on the simulation experiment in MATLAB environment to prove the superiority. On this basis, the load balancing strategy of the system is designed.
【學位授予單位】：廣東工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2012
【分類號】：TP391.41;TP333
【引證文獻】
相關碩士學位論文 前1條
1 陳聰;基于云存儲的視頻監(jiān)控平臺[D];華南理工大學;2012年
，本文編號：2493525
本文鏈接：http://sikaile.net/kejilunwen/jisuanjikexuelunwen/2493525.html
最近更新
教材專著
