Auction Method and its Performance in a Dynamic Bandwidth Allocation Service

Auction Method and its Performance in a Dynamic Bandwidth Allocation Service
Eiji TAKAHASHI and Yoshiaki TANAKA
Global Information and Telecommunication Institute, Waseda Unviersity 5th Floor. Nishi-Waseda Bldg., 1-21-1 Nishi-Waseda, Shinjuku-ku, Tokyo, 169-0051 Japan Tel: +81 3 5286 3831 Fax: +81 3 5286 3832 E-mail: etaka@giti.waseda.ac.jp, ytanaka@mn.waseda.ac.jp Abstract: In leased line services used by ISPs (Internet Service Providers) the bandwidth is fixed, but the traffic changes dynamically. To address this, we propose an auction method for the dynamic allocation of bandwidth to ISPs sharing backbone networks. By the method, backbone networks can be used effectively as each ISP is able to secure bandwidth according to its own policy. The Internet users can also be expected to receive good services, as it has been made possible for them to get all information about all ISPs, such as access fee and QoS provided, and to select their congenial ISPs among all ISPs according to this information. In this paper, we compare a dynamic bandwidth allocation service (dynamic allocation of bandwidth to ISPs) with a leased line service (fixed allocation of bandwidth to ISPs) by using the users’ utility to estimate the effectiveness of the proposed method. Key Words: ISP, bandwidth allocation, leased line, backbone networks, pricing I. Introduction charge is close to the low price end when the network is not congested; (2) the packet charge is close to the high price end when the network is congested. This method, however, poses some problems. One problem with this method is the variability of packet charges. If these packets charges are decided dynamically according to congestion states in the networks, the final price and the final provided QoS cannot be informed to any user at the start of packet sending. In this respect, this method is not user-friendly. A second problem with this method involves the accounting and billing costs. Costs become astronomical if detailed accounts are kept for every packet sent. Another problem with this method is uncertainty of the users’ benefit function. For a central planner to calculate efficient prices, the planner must know this users’ benefit function. However, almost no useful data exists on how Internet users shift their usage patterns over time to respond to packets charges [1]. These problems can be solved by replacing users with ISPs. This paper shows an auction method for the dynamic allocation of bandwidth to ISPs sharing backbone networks. By allocating resources dynamically to ISPs, the backbone networks can be used effectively as each ISP is able to secure bandwidth according to its own policy. Internet users can also be expected to receive good services, as it has been made possible for them to get all information about all ISPs, such as access fee and QoS provided, and to select their congenial ISPs among all ISPs according to this information. Auction is one approach to allocate resources fairly and efficiently. In the method, ISPs make bids for bandwidth for backbone networks. Each bid price is decided on the basis of the need to get bandwidth. This need is derived from several factors. They include the users’ traffic, the estimated income, etc. This paper describes how a network operator can allocate bandwidth dynamically to ISPs by using an English auction approach. Then, we compare a dynamic bandwidth allocation service (dynamic allocation of bandwidth to ISPs) with a leased line service (fixed allocation of bandwidth to ISPs) by using the users’ utility to estimate the effectiveness of the proposed method.
ቤተ መጻሕፍቲ ባይዱ
On the Internet, where a wide variety of services are provided by a number of competing ISPs, there has come to be a need recently to use network resources effectively to achieve economical and high-quality networks. In current leased circuit services used by ISPs, the bandwidth is fixed, but the traffic changes dynamically. Accordingly, the Internet has been known to become congested at times, with simply too much traffic for routers and lines to handle. At present, the way for the Internet to deal with congestion is to drop packets, making it necessary for some information to be resent by the application, or to delay traffic. These solutions unfortunately impose external social costs [1]. Therefore, a method based on microeconomics has been considered to control Internet congestion, and to use network resources effectively [1, 2, 3, 4, 5, 6]. In this method, the traffic is controlled by using statedependent pricing. More specifically, in order to collect the above external social costs, and moreover, to avoid building extra capacity to meet peak usage demands, price incentives are provided in an effort to shift some of the demands from peak to non-peak periods. This method has the following structures: (1) the packet