| Summary: | 碩士 === 國立嘉義大學 === 運輸與物流工程研究所 === 94 === The demands for Internet service have grown tremendously over the years as these are applied by various organizations such as business establishments,educational institutions, government organizations and individuals for email,web browsing, data exchange, and so on.
Due to the requirement for powerful Internet service like web browsing with quick response times,real time interactive games, quick downloads of large files, rapid communication,interactive multimedia communication, and video communication,there are more and more countries that have developed admittance to high speed or broadband connections.With the evolution of Internet application, Internet users depend on Internet Service Providers (ISPs) to provideimmediate and always-on broadband access to the Internet and to satisfy the users' needs.
Under a competitive market, ISPs must provide a high level of service to their subscribers since reliable access to the Internet is significant for many users.The demands for Internet service are expected to increase in the near future. However,from the ISPs point of view,the most critical considerations for these companies are satisfying the demand of customers for various services and increasing earnings at the same time. Therefore,ISPs have to enhance their telecommunications infrastructure to satisfy future demands.
This study deals with the multi-firm, multi-period, and multi-commodity bandwidth resources investment problems.
Two cases are discussed. In both cases, we assume the budget in each period is fixed,demand increases over time and follows a known exponential distribution,and the quality of service for each Internet commodity is predetermined.Pricing is an exogenous variable in the first model while it is a decision variable in the second model. The purpose of these models is to maximize the present total profit through determining the investment strategies. These models are formulated as constrained nonlinear integer programming.
The solution of model 1 can be optimally solved by LINGO 9.0 software.We developed a hybrid genetic algothem to solve the second model.A series of numerical examples were presented to compare the efficiency and quality of the proposed approach with LINGO 9.0 software.Numerical examples indicate that the hybrid genetic algothem is very efficient computationally.
|
|---|
