Examining the Wireless Commons

Abstract

Debates about the efcient use of radio spectrum are beset by a fundamental conundrum: although unlicensed spectrum has fostered tremendous innovation, is it plagued by a potential tragedy of the commons in which unfettered access renders it useless. There is disagreement if such tragedies occur and what, if anything can be done in technology and policy to delay or defuse such an effect. Surprisingly, given the heat of the debates, current evidence of a tragedy in the wireless commons is mostly anecdotal. Although there are limitations to such an approach, it is easiest to discuss the presence and remedy of a tragedy for a particular technology. For many reasons, we will use the widely deployed 802.11 wireless networking equipment as our vehicle for exploring the commons. In this paper, we use measurements of wireless networks and show that a common metric that leads people to believe there is a tragedy really demonstrates nothing of the sort. We also show how system architecture determines how rapidly a tragedy occurs. The fundamental questions we seek to address include: 1) what defines a tragedy in unlicensed spectrum and 2) how serious is the problem? An additional desire of this work is to clearly distinguish between the tragedies that occur as a result of radio frequency interference and those that occur because of the design of the protocols associated with specific technologies. We also seek to capture this influence of architectural choices on the performance of systems in the unlicensed bands. Defining what constitutes a tragedy is more complex than it might first appear. The answer includes such dynamic factors as: the density of the devices, the environmental conditions, the usage patterns, the application demands, the protocols employed and the existence of other interferers. Modeling the problem is also difficult in that it requires combining theoretical capacity models with geographic information systems and other dynamic factors in a way that connects micro and macro models into a realistic depiction of the problem.