Connected Vehicles for Internet Access: Deployment and Spectrum Policies

Internet traffic from mobile users has been growing sharply. To meet the needs of those
users, it is important to expand capacity of networks that provide Internet access in cost effective
ways. This capacity has traditionally been provided by cellular networks. However,
expanding the capacity of those networks alone may not be the most cost-effective way to meet
the present and future growth of mobile Internet under some circumstances. In this dissertation,
we show that networks of connected vehicles can be an important way to complement the
capacity of cellular networks to provide mobile Internet access under several scenarios.
Connected vehicles may soon be widely deployed, forming mesh networks of short-range
connections among vehicles and between vehicles and roadside infrastructure. These
connections are collectively referred to as vehicle-to-everything, or V2X. Deployment of
connected vehicles and infrastructure is primarily intended to enhance road safety, and the U.S.
Department of Transportation has recently proposed a mandate of V2X devices in vehicles
using Dedicated Short Range Communications (DSRC) technology. Other applications are also
envisioned that include Internet access in vehicles connecting to roadside infrastructure serving
as gateways to the Internet.
In this work, we find that V2X-based networks are more cost-effective than cellular to
provide Internet access, in scenarios which DSRC devices are mandated in vehicles to enhance
road safety. This is true initially for densely populated urban areas, but over time V2X-based
networks would be cost-effective in less populated areas as well, as long as Internet traffic or
penetration of V2X devices grow as expected.
Local and state governments are expected to deploy roadside infrastructure for safety
applications. If that infrastructure is shared with Internet Service Providers for a fee, then V2XABSTRACT based networks are cost-effective in locations with even lower population densities than the
locations where it is cost-effective to deploy infrastructure for Internet access only. Moreover,
the sharing fee could help governments save in infrastructure costs. We find the pricing
strategies that maximize either cost-effectiveness or government savings. We estimate that
governments could save about one-fifth of the total cost to deploy safety infrastructure
nationwide in the U.S., if fees are set to maximize government savings. Although we find that
these prices may differ from the pricing strategy that maximizes cost-effectiveness, maximizing
government savings results in near-optimal cost-effectiveness.
The U.S. Federal Communications Commission has allocated 75 MHz of spectrum to be
used exclusively by DSRC devices, and it has been hotly debated whether all or part of that
bandwidth should be shared with unlicensed devices. We find that it is highly efficient to share
any spectrum allocated to V2X communications beyond the portion of that spectrum that is
needed for safety-critical DSRC messages. V2X and unlicensed devices require up to 50% less
bandwidth on shared spectrum to achieve given throughputs, compared to V2X and unlicensed
devices using separate bands. We conclude that the spectrum available for V2X should be
maintained or increased, as long as much of that spectrum is shared with non-V2X devices.
Conclusions are derived from an engineering-economic approach, in which part of the
assumptions are based on data from a citywide deployment of connected vehicles in Portugal.
The data is used in a detailed and realistic packet-level simulation model of V2X-based
networks used to provide Internet access with DSRC technology. In some scenarios, the
simulation also includes unlicensed devices using Wi-Fi technology. The results of the network
simulation are then fed into engineering-economic models to compare costs of V2X-based
networks with costs of macrocellular networks to carry given amounts of Internet traffic, and to
estimate other measures such as government revenues and spectrum usage. Those measures
help inform decisions about where and when to deploy V2X-based networks, decisions about whether and how to promote public-private partnerships to deploy V2X infrastructure, and
decisions about sharing spectrum used for V2X communications with non-V2X devices.