| The video services which based on peer-to-peer technique can serve a large number of users simultaneously with a small value of bandwidth. But with the survey, the costs of server bandwidth occupy half of the total costs, which the video service providers pay to ISPs. And the bandwidth utilization rate is less than20%for more than50%time. At the same time, with more and more users with mobile, PDA et al. join and watch the videos, their low capabilities of storage and bandwidth are harmful to the resource sharing. So how to improve the efficiency of resource sharing, decrease the bandwidth costs and increase the bandwidth utilization rate are the attention problems.With the rise and application of cloud, renting cloud platform to join the video services can decrease the investment and maintenance costs of servers. Considering the costs which the video service providers have been invested, it is an effective method to hybrid local servers and cloud platform to form cloud-assisted VoD system. It can decrease the costs of bandwidth and increase the bandwidth utilization rate. But in this system, questions must be solved, which are shown in the following:(1) How to share resources effectively with the conditions that more and more low-capability nodes are in the system?(2) How to judge the user demands of each video channel? Because the videos with high user demands must be migrated to cloud, then cloud can participate in distributing the demands.(3) How to request the resources and how to determine the migration strategies in the system with only one cloud platform? So it can achieve the purpose of decreasing costs and inproving the QoS of users.(4) How to further decrease the costs with the methods of distributing videos to cloud platform? In order to solve the problems metioned above, and with the essence of P2P techenique and the characters of renting and paying for cloud platform, we study deeply in decreasing the costs of video service providers. The major contributios are as follows.1. We present an efficient server bandwidth costs decreased mechanism towards low-capability nodes in cloud-assisted P2P-VoD system. We propose the NCDLT, which is based on charaters of nodes with poor storage and bandwidth. It is composed by three parts, which are neighbor selection algorithm based on buffer, data chunks selection algorithm and distribution taxation algorithm. Compared with BPB, the proposed mechanism can obtain lower bandwidth costs. And at the same time, it can improve the QoS of users.2. We propose a prediction mechanism which can forcast the user demands in days in P2P-VOD system. It quantizates the user demands of each video in future. We present the SBDP mechanism which is based on the banlance of uploading and downloading in P2P system. It computes the user demands of users according to predicting online population, peers’ uploading value and nodes’ downloading value. Combining with the DFPP, they can provide the basis which videos will migrate to the cloud platform.3. We present a video migration strategy based on bandwidth reservation in cloud-assisted VoD system with one cloud. It can improve the QoS of users with the condition that the costs increase a little. We propose the VMSBR which is bansed on the character that the "pay-as-you-go" can cause some delays. This method is based on the prediction of user demands. It optimizes the reservation bandwidth and designs the migration strategies according to the bandwidth changes of servers in different time in a day. Compared to the smart strategy, our method can decrease1%user rejection rate with the increasment of0.5%costs.4. We present a videos distribution method in cloud-assisted VoD system with more than one cloud platform. We propose the LBAS methord based on disaster of single cloud and the competition principle of market. It is based on auction and low-price win, and restriction of coalition. It can distribute user demands to different cloud platform. Compared to VMC, our method can further decrease the10%costs. |
PDF Full Text Request