Wednesday, March 6, 2013

Network Economics and the Internet

In 1993, my first book, Network Economics: A Variational Inequality Approach, was published as the first volume of  the ACE series: Advances in Computational Economics. The second and revised edition of this book came out in 1999.  My 10th book, Networks Against Time, was just published by Springer.

Since the publication of my Network Economics book, there have been many synergies associated with the modeling of the Internet and other network systems, including decentralized transportation networks, with even the Braess paradox, which was identified in 1968,  being relevant also to the Internet.

What I find amazing, and actually quite exciting,  is that some of the language associated with the Internet is similar to that which we are encountering and using in supply chain networks -- from service providers to transport network providers, to start.

I have been involved, since last year, in a truly innovative and challenging project: funded by the National Science Foundation (NSF) entitled Collaborative Research: Network Innovation Through Choice. The investigators on this project are: Dr. Tilman Wolf (lead PI) and Dr. Anna Nagurney (Co-PI) - UMass Amherst; Dr. Jim Griffioen and Dr. Ken Calvert - University of Kentucky; Dr. Ilia Baldine - RENCI at UNC, and Dr. Rudra Dutta and Dr. George Rouskas - NCSU.

We have regular teleconferences during which we discuss our research and brainstorm and it is fascinating to be working with such top electrical and computer engineers and computer scientists. Together, we generate many good ideas and the discussions are always very stimulating. There are several doctoral students also involved in this multiyear project. I am supporting, through this grant, one doctoral student in Management Science at the Isenberg School of Management at UMass.

The project that we are working on we are calling ChoiceNet and we have two papers with all 7 of us as co-authors: our SIGCOMM paper, "Choice as a Principle in Network Architecture," and an invited paper, "ChoiceNet: Network Innovation Through Choice," that Dr. George Rouskas will be presenting in France next month at the Optical Network Design & Modeling Conference.

As we note in the latter paper, The Internet has been amazingly robust to changes in both technology and usage over the past three decades. A large reason for this success has been the fact that users are able to create and deploy a wide range of applications, devices, and services on end-systems around the edge of the network. In addition to enabling the Internet to support new uses that were not envisioned by its creators, this has also led to a vibrant economic environment for innovation at the application layer.

The many successful Internet companies that have arisen in the last two decades testify to the ability of a platform on which novel ideas can be developed, tested, and adopted to lead to value creation. Ironically, the same paradigm (i.e., the ability to create new functionality and let users choose winners and losers) is not presently supported inside the network. It is widely agreed that the current Internet architecture inadequately supports adaptation in the data and control planes. 

In the early days of the Internet, innovations were deployed in the core by consensus among a small community of researchers and operators. With today’s dramatically larger community, consensus is more difficult to achieve and innovation is for the most part limited to the edge. This limitation inhibits the development and deployment of new networking services, protocols, security designs, management frameworks, and other components that are essential to support the increasingly diverse systems, applications, and communication paradigms of the next-generation Internet.

We are hoping through a vision for the Internet to offer more choices and this entails the crucial economy plane. Users should not be stuck with contracts over extended time horizons and should be able to dynamically select service and transport network providers.

The underlying network economics issues from contracts to game theory models of competition from Cournot and Nash to Bertrand are very intriguing and we have been developing computable game theory models using variational inequality theory and projected dynamical systems theory. Drawing on some results in my first book we are also designing and constructing extensions and pushing the methodological foundations forward.

For more information on the NSF Future Internet Architecture (FIA) projects click here.