Innovative Applications of O.R.Cooperative interconnection settlement among ISPs through NAP
Introduction
Internet users perceive the Internet as seamless and global, while behind the scene there exist many individual networks connected with each other instead of a sole massive network. Internet Service Providers (ISPs) connect their networks to each other by interconnection arrangements in order to enable communication among end-users from different networks. There are three interconnection modes: NAP (Network Access Point) peering, private peering and transit. When an ISP decides to connect itself to other networks, it has to consider the impact of the interconnection arrangements on its cost and profit and decide its interconnection strategy.
In China, ISPs, such as China Netcom and China Telecom, interconnect with each other mainly through private peering and NAP peering. Currently there are only ten national NAPs located at major cities such as Beijing, Shanghai and Guangzhou etc., and the quality of communication between networks and the data transmission efficiency across different regions are relatively low due to the limited number of NAPs. To facilitate data exchange and extenuate traffic congestion at national NAP, local NAPs have been built successively in several cities since 2004. This kind of local Internet exchange point has been proven beneficial to local Internet ecosystems, especially in developing countries and regions such as Latin America (Galperín, 2016, Weller and Woodcock, 2013). Similarly, the construction of these local NAPs in China has greatly reduced the interconnection costs and improved the network response speed, data exchange quality and safety.
However, new problems arose as local NAPs were put into use. For example, Shanghai Network Access Point SHNAP (SHNAP) initially adopted a non-settlement rule for the data exchange between its member ISPs. Under the non-settlement rule, ISPs do not charge each other for data exchange. Since SHNAP can save great money for ISPs for interconnection, it attracted 16 ISPs to connect to it and the data exchange volume increased very fast in the first few years. The non-settlement rule, however, began to show its inefficiency after several years of implementation, as it can hardly reflect the cost and revenue to peer in SHNAP for each ISP. As a result, it caused an imbalance of profit allocation among member ISPs, and ISPs, especially those large ones, had no incentive to connect to SHNAP and would rather peer with other ISPs privately. Similar problem in Argentina has also been documented by Galperín, (2016). Furthermore, ISPs who have already connected to SHNAP were reluctant to invest in capacity to improve the interconnection quality. At this stage, the introduction of a rational settlement rule, which allows fair profit allocation and stimulates ISPs to exchange traffic in local NAP, becomes the key to promote local NAP peering and enhance the value of local network.
Our study is developed against the general background of SHNAP. SHNAP wants to find a settlement rule that can make ISPs’ selfish behavior results in globally optimal interconnecting decisions, i.e., all the local ISPs choose to interconnect with each other through SHNAP and make jointly optimal decisions. To achieve this goal, the profit allocation induced by this settlement rule must be fair in profit distribution and can bring more profit to each ISP. As NAPs are continually being set up all over the world for both regional interconnection and global interconnection (such as London Internet Exchange which has members from 40 countries), this study can provide valuable managerial insights on how to improve NAP peering efficiency not only for SHNAP, but also for practitioners worldwide.
In this study, we first build an analytical model of individual ISP's demand and profit and develop a cooperative game framework to analyze the different profit allocation rules. We show that interconnection with other ISPs can bring extra profit to ISPs. We further check whether the allocation rules, i.e., non-settlement profit allocation, the Shapley-value based profit allocation and the Characterized Profit Allocation we propose, can encourage ISPs to interconnect with each other and, at the same time, demonstrate fairness in profit distribution. Results show that the non-settlement allocation, which exerts no side payments on ISPs, is not in the core of the game, and does not preserve fairness either. The Shapley-value based profit allocation, which is known as fair allocation rule, on the other hand, is in the core. As Shapley-value based allocation requires complex computation and is difficult for ISPs to understand, we propose a new allocation rule, i.e., the Characterized Profit Allocation, which is easy to interpret. Our analysis shows that CPA is in the core and preserves fairness, and that it performs well in different situations. Numerical experiments are conducted to support our conclusions from theoretical analysis.
The rest of the paper is organized as follows. Section 2 reviews the related literature studying Internet interconnection, especially those who endeavored to approach a cooperative settlement rule. In Section 3, we set up the baseline model, including demand function and profit function. Section 4 builds the cooperative game model to analyze the impact of different profit allocation rules on ISPs’ profit, and proposes a settlement rule to implement the Characterized Profit Allocation. Section 5 extends the basic model in two directions: consider the interconnection quality choice of ISPs and introduce competition in the market, and examine if the proposed settlement rule can encourage ISPs to connect to NAP under these two scenarios. In this section we also analyze a model with linear network externality. Numerical experiments are conducted in Section 6. Section 7 summarizes the findings and discusses future research directions.
Section snippets
Related literature
Under non-cooperative game analysis framework, there is a large body of literature studying ISP interconnection strategies, including determining compatibility and access charges. Cremer, Rey, and Tirole (2000) develop their research on basis of Katz and Shapiro's model of network externalities (Katz & Shapiro, 1985) and study the strategies of Internet backbone providers. They use a Cournot-cum-installed-bases model and show that compared with small Internet Backbone Provider (IBP), larger IBP
The model
In this section, we describe our basic model and assumptions, and analyze the benchmark system where each ISP does not interconnect with others.
We consider a set of ISPs . Each ISP is characterized by two parameters. The first is the intrinsic demand potential Di, which is related to the coverage area of ISPi and is considered to be exogenous. The second parameter is the installed network size ei, which can be composed of the number of installed end-users and the richness
Settlement analysis in a cooperative game framework
When an ISP interconnects with other ISPs (see Fig. 1), its subscribers can visit networks interconnected with it. On one hand, more consumers will be attracted to the ISP since they can visit more resources, and the ISP can generate more revenue by serving more end-users. On the other hand, interconnection will also incur an extra cost, as the data transmission cost increases with the increasing number of end-users and the expanded associated network size.
With the demand function defined in
Incorporating quality decision in interconnection settlement
In practice, ISPs with asymmetric network sizes usually have asymmetric incentives to provide interconnection quality, which will determine how useful and efficient the interconnection is. To address this issue, we extend our model to consider ISPs’ interconnection quality decision along with pricing decision. Some scholars have looked into the similar price and quality-based competition problem in the interconnection market. Matsubayashi and Yamada (2008) study how the asymmetry in consumer
Numerical experiments
From Sections 6.1 to 6.3 numerical experiments are conducted to illustrate the effectiveness of CPA in basic model. In Section 6.4, we provide numerical evidence of the effectiveness of the proposed settlement rule s in a competitive market with two players.
Conclusion
ISPs in a specific Internet market exchanging data through NAP can give rise to the welfare of the whole society. For the Internet users, they can visit more resources with shorter delay. For ISPs, as long as there is a rational settlement rule, exchanging data with other ISPs through NAP will also increase their profit. If there is no such settlement rule, ISPs’ profit could be hindered, and would rather choose to operate alone or form small coalitions with some of the other ISPs. In our
Acknowledgment
We thank the editor and three anonymous reviewers for improving this work with their insightful comments and criticisms. This work was supported by NSFC under grants No. U1509221, No. 11301479 and No. 71571160.
References (39)
- et al.
A simple game-theoretic analysis of peering and transit contracting among Internet service providers
Telecommunications Policy
(2008) - et al.
A solution concept for network games: The role of multilateral interactions
European Journal of Operational Research, 243
(2015) - et al.
Competition and compatibility among Internet service providers
Information Economics and Policy
(2001) - et al.
Do internet incumbents choose low interconnection quality?
Information Economics and Policy
(2005) Localizing Internet infrastructure: Cooperative peering in Latin America
Telematics and Informatics
(2016)- et al.
Interconnection and competition among asymmetric networks in the Internet backbone market
Information Economics and Policy
(2008) - et al.
Pooling of spare parts between multiple users: How to share the benefits?
European Journal of Operational Research
(2014) - et al.
Core-based cost allocation in the cooperative traveling salesman problem
European Journal of Operational Research
(2016) - et al.
QoS commitment between vertically integrated autonomous systems
European Journal of Operational Research
(2011) Asymmetric access pricing in the Internet backbone market
Economics Letters
(2011)
Cooperative game theory approach to allocating benefits of horizontal cooperation
European Journal of Operational Research
A cost allocation problem arising in hub–spoke network systems
European Journal of Operational Research
A note on price and quality competition between asymmetric firms
European Journal of Operational Research
Connectivity in the commercial Internet
The Journal of Industrial Economics
Cost allocation of capacity investment games
Naval Research Logistics
Can bilateral ISP peering lead to network-wide cooperative settlement
Pricing and revenue sharing strategies for internet service providers
Revenue Sharing in Airline Alliances
Management Science
Cited by (10)
Quality competition among internet service providers
2023, Performance EvaluationA note on “Cooperative interconnection settlement among ISPs through NAP”
2017, European Journal of Operational ResearchCitation Excerpt :In a recent paper in this journal, Zhou, Pu, Dai, and Jin (2017) examine how Internet Service Providers (ISPs) that interconnect through Network Access Points (NAPs) can most effectively allocate total profit.
Sustainable supply chain management with pricing, greening and governmental tariffs determining strategies: A game-theoretic approach
2017, Computers and Industrial EngineeringResearch on the selection of interconnection modes between asymmetric internet backbones
2022, International Journal of Information and Communication Technology