[Back]


Talks and Poster Presentations (with Proceedings-Entry):

T. Blazek, C. Mecklenbräuker:
"Low Complexity SNR-Based Packet Level Burst-Error Model for Vehicular Ad-Hoc Networks";
Talk: 2018 IEEE 88th Vehicular Technology Conference (VTC2018-Fall), Chicago, IL, USA; 08-27-2018 - 08-30-2018; in: "2018 IEEE 88th Vehicular Technology Conference (VTC2018-Fall)", (2018), 1 - 5.



English abstract:
Network simulators are a crucial tool for evaluat- ing the performance of Vehicular Ad-Hoc Network (VANET) protocols. They allow the assessment of the scalability and the influence of geometric topologies. However, typical network simulators such as NS-3 or OMNET++ often resort to overly simplified packet error models, limiting the validity of the results. Measurements have shown that burst-error patterns are characteristic to VANETs, which are not modeled using the common approaches. In this contribution, we develop a low- complexity burst error model that is parameterized by the Signal- to-Noise Ratio (SNR). We adapt an approach based on the Gilbert-Elliot Markov model, which allows to model first order burst properties while retaining low complexity. Furthermore, we define three performance indicators, overall error probability as well as burst error and burst success probability, and use a multi-feature information bottleneck to find the optimal SNR quantization in the mutual information sense. Based on this, we present Gilbert-Elliot model fits that are easily implementable and demonstrate that low-level SNR quantizations of 4-9 intervals are sufficient to capture the statistics in the MSE sense.

German abstract:
Network simulators are a crucial tool for evaluat- ing the performance of Vehicular Ad-Hoc Network (VANET) protocols. They allow the assessment of the scalability and the influence of geometric topologies. However, typical network simulators such as NS-3 or OMNET++ often resort to overly simplified packet error models, limiting the validity of the results. Measurements have shown that burst-error patterns are characteristic to VANETs, which are not modeled using the common approaches. In this contribution, we develop a low- complexity burst error model that is parameterized by the Signal- to-Noise Ratio (SNR). We adapt an approach based on the Gilbert-Elliot Markov model, which allows to model first order burst properties while retaining low complexity. Furthermore, we define three performance indicators, overall error probability as well as burst error and burst success probability, and use a multi-feature information bottleneck to find the optimal SNR quantization in the mutual information sense. Based on this, we present Gilbert-Elliot model fits that are easily implementable and demonstrate that low-level SNR quantizations of 4-9 intervals are sufficient to capture the statistics in the MSE sense.

Keywords:
Intelligent Transportation Systems, Network Simulations, VANETs


Electronic version of the publication:
https://publik.tuwien.ac.at/files/publik_270356.pdf


Created from the Publication Database of the Vienna University of Technology.