A multiple period bandwidth allocation model with overbooking consideration

• Main Authors: Chieh-Ming Huang, 黃介民
• Other Authors: Peng-Sheng You
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/48006848538768091782

碩士 === 國立嘉義大學 === 運輸與物流工程研究所 === 94 === With the evolution of telecommunications infrastructure, the internet has been successfully used in various fields. The application involves people that can access to interactive multimedia communication and video communication for their conference. People can also use it to access web mail and web page, as well as use it as a platform for internet telephone. Due to the powerful application, demands for internet service significantly increase. In practice, the bandwidth resource required to support the quality of the internet services are different. For example, the bandwidth resource required by video conference is much higher than that of email. In addition, the bandwidth resource required by VOIP is larger than that of web access. Therefore, there are distinctive bandwidth requirements given according to the needs of internet users. Accordingly, the internet service providers frequently offer a variety of internet commodities to serve their customers. Each internet commodity has a unique bandwidth allocation and service quality. While the bandwidth resources are divided into numerous definite internet commodities for sale, the internet providers face the problem of allocating their bandwidth resource to distinct internet service commodity. In this paper, we refer to this problem as bandwidth resource allocation problem. Since internet users do not always access internet, the number of online users is usually smaller than the number of users (i.e., offline users plus online users). To exploit their bandwidth resource, the internet providers mostly employ the overbooking strategy to improve their revenues. However, problems associated with management arise in how to set the overbooking limit for every internet commodity. This paper developed two mathematical models to deal with this problem. The first and the second models are referred to as a static single commodity internet bandwidth overbooking model and a multi-period-multi-commodity internet bandwidth allocation model. In the first model, we allowed the booking limits to be real numbers. This paper provided conditions for achieving the most favorable decision in which the optimal overbooking limit is obtained by a nonlinear equation. The second model is a constrained nonlinear integer problem. In this model, the booking limits are assumed to be integers. The problem is of an NP hard problem. We provided an efficient computational approach to acquire a satisfied solution. The solutions obtained by the heuristic approach are compared with those found by the lingo software. Numerical results show that the proposed heuristic approach only require a small number of CPU time to attain confidential solutions.