Towards the Efficient Design of Vehicular Ad-Hoc Networks

Jared Dulmage

Advisor: Michael Fitz

Abstract:

Vehicular ad-hoc networks (VANETs) hold the potential to greatly improve the safety, efficiency, and comfort of automotive transportation. VANETs consist of both mobile, vehicular nodes and fixed road-side units all connected over wireless links. Potential applications include advanced collision warning, toll collection, and coordinated driverless vehicles. With public safety a priority, robust design and performance analysis of the end-toend system is critical.  Unfortunately, accurately modeling the protocol stack and channel for possibly thousands of nodes is computationally infeasible.

This dissertation presents four results that contribute to more efficient VANET simulation and simplify analysis. First, we present a software channel model that accurately models the time-variations characteristic of the highway environment. Second, we consider goodput, throughput accounting for packet errors, for a typical VANET scenario. Using a semi-analytic analysis, we relate the packet size to goodput given the channel coherence-time. This result could replace the computationally expensive physical layer and channel model in a VANET simulation of the full protocol stack.

The third result derives the approximate mutual information (MI) variance for PSAM-OFDM accounting for both training overhead and channel estimation error. Approximating the MI distribution as Gaussian, the result enables optimal bandwidth selection of PSAM-OFDM given the maximum excess channel delay to maximize goodput. The final result derives closed form bounds and a simple approximation for the MI variance of narrowband PSAM over a time-varying channel, highlighting critical parameters and facilitating design for outage capacity and goodput.

Biography:

Jared Dulmage received his B.S. in Electrical Engineering and Computer Science from Duke University. He received his M.S. in Electrical Engineering from the University of California, Los Angeles. He is currently working towards his Ph.D. degree in wireless communications under advisor Dr. Michael P. Fitz. His research interests include cross-layer design, channel modeling, and high-performance simulation.