Measurement-Based Burst-Error Performance Modeling for Cooperative Intelligent Transport Systems

Abstract

Safety applications for cooperative intelligent transport systems are limited in their performance by the latency of the communication more than by the achieved throughput. However, there exist few models at packet level that are able to capture the burstiness of the communication. We therefore introduce a packet error model that considers burst lengths through second-order statistics and mean packet errors. The foundation of our approach is the Gilbert-Elliot model, which is able to model not only the packet error rate, but also the burst durations of the packet errors, which we interpret in a nonstationary fashion. Based on this, we formulate maximum likelihood expressions for the time variant model fits, and then proceed to fit the parameters to extensive recorded measurements. We consider the fading statistics of the measured channel and the signal-to-noise ratio and present how they influence the channel burstiness. Our analysis demonstrates that the communication shows strong bursts at packet level, proving the demand for such models. The approach we demonstrate here remains of low computational complexity, allowing future employment in large-scale simulations.