Research On The Deployment Of Network Resources And Advertisement Insertion In Online Video System

Online video has become the mainstream industrial system on the Internet,occupying most of the network traffic.In this industrial system,content providers rent bandwidth and cache resources from the internet service provider to provide video content to end-users.Content providers' revenue consists of the users' subscription fee and advertising revenue by inserting video advertisements.To maximize the overall revenue,how to rent cache and bandwidth,and how to insert video advertisements are important issues that content providers focus on.Based on the existing research,in this paper' we study the network resource renting strategy and the video advertisement insertion strategy from the above two perspectives respectively.First,we study the optimal budget allocation problem of content providers.According to the different contributions of cache and bandwidth to user experience,the strategy of budget allocation to rent cache and bandwidth is optimized.The objective is to maximize the user subscription revenue of content providers with the consideration of user watching experience.Secondly,we optimize the video advertisement insertion problem.According to the different effects of ad insertion proportion and position on the probability of users leaving,we analyze the impact of ad insertion strategy on the overall advertising revenue of content providers,to determine the best advertisement insertion strategy.In this paper,we model the above two problems as corresponding optimization problems.Then we find out the impact of various factors in the system on the revenue of content providers.We analyze and solve the problems and find the optimal budget allocation strategy and the optimal advertisement insertion strategy.Specifically,the main contributions of the dissertation include the following two aspects:(1)Aiming at the problem of budget allocation,we establish a budget-constrained resource optimization allocation model.The model is based on two correlations in the system.One is the non-linear negative correlation between network resources(bandwidth and cache capacity,respectively)and video transmission delay.The other one is the negative correlation between user required video data volume and the transmission delay.The objective of the model is to find out the best bandwidth and cache resource deployment strategy to maximize the overall revenue of content providers.In this paper,based on the positive correlation between the optimal cache and the optimal bandwidth,we design the feasible-plane algorithm to find out the best allocation strategy.The numerical results verify the effectiveness of the algorithm under different system parameter settings.(2)Aiming at the problem of ad insertion,we establish an ad insertion model to find the best ad insertion proportion and position.This model is based on the natural departure probability of the user watching behavior and combined with the influence of the ad on the users' departure probability.The objective of this model is to maximize the advertising revenue of content providers.In this paper,we derive the theoretical optimal solution for the advertising proportion under a given insertion position,and analyze the trend of advertising revenue under the combined effect of the two factors:insertion proportion and position.Numerical analysis shows the best ads insertion strategy under combined effect of the given user's natural departure probability and the probability of users leaving due to ads insertion.