Study On Several Vital Problems Of Pricing Scheme Based On Microeconomic For DiffServ

"Best effort"Internet is enough to support services such as web browsing, e-mail, file transfer, but it is not sufficient for real-time services such as audio and video which require a much better quality of service (QoS) model. DiffServ is one of schemes providing QoS; there are two methods to implement: non-economic and economic resource allocation method. By the non-economic method, when allocating network resource, the method don't consider the willingness to pay and the value of the resource to one user, which results in the inefficient allocation. In contrast, the goal of the economic method is maximizing the total utility by allocating resource among users and demand. Moreover, by users'self management, the congestion is auto controlled. The pricing scheme based on the theory of supply and demand was introduced by Kelly, the link prices to the services (such as bandwidth) is adjusted according to the supply and the demand. Based on existing research results for pricing, this paper investigates that how the games between ISPs and users influence on the network market, the convergence of the first-order update law, that the feedforward doesn't function properly, analyzes the influence of the principal-agent game on ISP's(Internet Service Provider, ISP) and users'utility, puts forward the second-order update law and a novel algorithm to enhance the control function of the feedforward. The game theory is introduced into the pricing scheme based on microeconomic to analyze the influence of the outcomes that strategic interactions among rational players produce, the players include the users, ISPs. The principal-agent game in static pricing scheme for priority services is analyzed to illustrate the influence of the game among players. Because there exists incomplete information associated with ISP's behaviors between the principal (ISP) and the agents (users), the principal's aim is to extract the maximum possible surplus from the agent, so it is possible that he forges. In response to forging, users take actions, if he detects that ISP is forging, he won't pay his bill. The influence of that game on ISP and user's utility was analyzed and simulated. The first-order update law is commonly used to update the estimate price. Only under strict condition the model can converge. When the condition is not satisfied, the link price oscillates wildly. The novel algorithm is proposed to enhance the convergence of the model. Now, under mild condition, the model converges, even if it can't converge, the link price oscillates slightly around the equilibrium price. For feedforward flow control pricing scheme, in order to choose his demand to maximize his utility, one user firstly estimates the price in next time slot, then compares his expect payoff with the willingness to pay. Through predicting the influence of his behavior, the user can adapt his demand to the relation between the total supply and the total demand in next time slot. So if the prediction of the decision's influence on the link state is inaccurate, the user's demand can't reflect the real circumstance in next slot, the function of the feedforward is weakened. This affects the convergence of the system. Novel method is presented to address to estimate the demand. It is proved that the dynamic estimate method is more suitable: when predicting that the total demand excesses the supply, the user reduces his demand, when that the supply excesses the total demand, the user consumes more service.