The Future of Wireless Resource Management: Bootstrapping and Automated Negotiation
The history of wireless communications can be told through the history of resource management. In this talk, I will describe the ongoing evolution from static to dynamic resource management and the concurrent evolution from homogeneous to heterogeneous networks.
From this description, I will draw two themes for the future: (1) a need for a pragmatic, bootstrapped approach to dynamic resource assignment, and (2) a demand for robust approaches to automated negotiation between radio agents. Along the first theme, I will show how unbounded dynamicism leads to failure; instead, we need pragmatically engineered systems that can bootstrap wireless communications from simple beginnings to complex sharing schemes.
In particular, I will describe a proposed channel assignment scheme for cognitive radio networks that balances the need for topology adaptation to maximize flow rate and the need for a stable baseline topology to support network connectivity. We focus on networks in which nodes are equipped with multiple radios or transceivers, each of which can be assigned to a channel. First, we assign channels independently of traffic, to achieve basic network connectivity and support light loads such as control traffic, and, second, we dynamically assign channels to the remaining transceivers in response to traffic demand. We formulate the problem as a two-stage mixed integer linear program (MILP) and show that with this two-stage approach we can achieve performance comparable to a fully dynamic channel assignment scheme while preserving a static, connected topology.
In the second theme, I will then describe the necessity of automated negotiation in future systems, including the relationship between these negotiated dynamics and the bootstrapped approach discussed in the first theme. In particular, I will explain some of the tentative steps that we have taken in this direction, including applications of auction theory and coalition game theory.
Finally, I will introduce some potential tools and approaches for future research.