Problem Formulation
In this section, we propose a resource allocation among users, brokers, and media cloud. Firstly, the resource allocation framework is introduced. Then, the utility function of each player is defined. Finally, the structure of game is elaborated in detail.
Resource Allocation Framework
There are three parties in the resource allocation framework, which are media cloud, brokers, and users. Media cloud determines the price of cloud resource p, which denotes the unit of resource to be paid by brokers. Each broker i decides the amount of cloud resource Ei to buy, and then announces the price of the bought resource pi to users. Each user decides a broker to connect for acquiring media service and obtaining the satisfied QoE. Due to the social features of MSNs, users in the same community are friends who can communicate with each other. Thus, in the same community, each user can know the information of others’ connections and then compare the utility of his connection with others. If anyone’s utility is better than his, he can change his connection.
The choices of three parties can mutually affect the decision of each other. The higher price of resource is determined by a media cloud, the less cloud resource that the brokers may decide to purchase, even this broker having heavy tasks. At the same time, if one broker announces a higher price to users, the user will connect other broker who announces lower price.
To allow the coordination among three parties to choose the proper parameter for enhancing performances, we propose a resource allocation framework with four steps. Firstly, the media cloud determines the price of its cloud resource and broadcasts it to brokers, aiming to acquire the revenue. There is a tradeoff between the revenue and the price for media cloud, e.g., if the price is too high, the demand of cloud resource will be reduced, which will affect the overall revenue. Secondly, after receiving the price of resource, each broker can buy a certain size of cloud resource from the media cloud to satisfy the demand of users who connect to this broker.
Thirdly, each broker decides the price of resource to achieve revenue from users. At last, each user will select a broker for connection and send his task to the broker.
