Measurement Based Vehicle-to-Vehicle Multi-link Channel Modeling and Relaying Performance

Abstract

There has been intense research in vehicular communication in order to provide reliable low-latency vehicular communication links for developing intelligent transportation system (ITS). As one of the important properties, vehicle-to-vehicle (V2V) communication is learned to be inherently non-stationary due to the high mobility of both transmitter (TX) and receiver (RX). Therefore, the V2V system behavior is essentially different from previous mobile communication studies and needs to be understood.

For V2V wireless communication systems, it is crucial to model the vehicular channel accurately to evaluate the quality of the system level applications. Among all channel properties in a V2V system, the shadow fading (i.e. large scale fading, LSF) from other vehicles has a significant adverse impact on the system performance. One promising
approach to overcome this issue is by implementing multi-hop technology on the vehicular ad hoc network (VANETs). One goal of this thesis report is to implement relaying schemes on simulated Rician channel based on measurements to evaluate the performance of multi-hop technology in V2V systems. Two relaying schemes, Amplify-and-Forward (AF) and Decode-and-Forward (DF), are employed in the bit level simulation. The results of packet error rate (PER) are evaluated together with non-relaying situation for convoy and overtaking scenarios, respectively.

Furthermore, a statistic model is created to model the measured highway environment. Pathloss parameters and shadowing loss together with correlation coefficients are derived. Line-of-sight (LOS) and obstructed line-of-sight (OLOS) conditions are manually separated through on-board video. Each scenario has its own parameter set. Maximum likelihood estimation (MLE) is utilized on the pathloss model to compensate the biasing from the measurement hardware. Also the shadowing is modeled as correlated Gaussian and we derived the decorrelation distance from the auto-correlation function (ACF). The model is also validated against the measurements.

For an ad hoc network, the diversity schemes would be strongly affected by the multilink correlation. Only a few joint correlation studies for mobile ad hoc network have been made, but rarely for VANETs. The last goal of this report is to study the joint correlation on VANETs based on measurements for four-dimensional position joint correlation model where shadowing is affected by the vehicle distance. To be precise, we focus on the joint correlation of large scale fading affected by the distances between the two receiver vehicles under the same car obstruction. Finally, a stochastic model based on the sum of sinusoids approach is implemented.

Popular Abstract

The fifth generation wireless systems denotes the next major phase of mobile telecommunications standards and is expected to meet consumer demands by 2020.
One of the major approach is the vehicular ad hoc networks, which is a spontaneous creation of a wireless network for data exchange to the domain of vehicles.
As a key component of the intelligent transportation systems, it is extremely importance to model the vehicular propagation channel in order to meet the requirement of low latency and high reliability.

There are many parameters that could describe the channel characteristics. Among them all, the vehicular shadowing has a significant advise impact on the system performance which describes the signal fluctuation affected by an obstruction vehicle.
In order to study it, a measurement was designed and took place in Sweden, road Rv 40. Four Volvo cars were forming convoy and overtaking scenarios with equipped signal-transmit-receive devices. Three major works are done in this thesis project based on this measured highway scenario:

Firstly, a simulation is designed based on the multi-hop technology. A scenario is simulated where a source car sends packets to a destination car with the help of a relay car. All the packets are randomly generated with each containing proper coding and modulation to enhance the transmission quality as well as to check the success or failure of the transmission.
The power properties of the wireless channels between each car-link are captured from the the measurements. They are simulated with a vehicular-based distribution while each byte of the signal would experiences a vehicular channel more close to practice.
Two relaying schemes are implemented in the simulation with a different reaction at the relay car after receiving the signal from the source. The destination then combines the two signals from the source and relay. It decodes the signal and records the decoding results.
For each observation, the ratio of successfully transmission number and total transmission number is recorded as packet error rate. Eventually, the packet error rate performances of different schemes are compared and evaluated.

Secondly, based on whether the link between antennas are obstructed, two scenarios are separated manually by watching the on-board videos. After that, the signal penetration based on transmission distance and the signal fluctuation affected by large obstructions are modeled based on the individual scenario. An advanced estimation method is employed during the modeling of the signal penetration to efficiently include the lost packet information. As for the modeling of the signal fluctuation, an extended distribution is used to describe the facts that, when a transmission link is obstructed by another vehicle, it normally remains obstructed for a certain amount of time. Eventually, channel power can be regenerated based on the model containing the measured channel properties.

At last, a model is created to describe the joint effects on the signal fluctuation based on the vehicles' movement and the distances between two receive cars whose signals are obstructed by the same vehicle.
The vehicular shadowing following a certain distribution is approximately represented by the sum of many sinusoid waves with random phases and chosen frequencies. The frequencies are generated based on the power and joint correlation properties of the measured channel based on the two effects.

• author: Chen, Xi
• supervisor: Fredrik Tufvesson
• organization: Department of Electrical and Information Technology
• course: EITM02 20152
• year: 2015
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• keywords: V2V, wireless, channel model, multilink, relaying, sum of sinusoids, correlated Gaussian, maximum likelihood estimation
• report number: LU/LTH-EIT 2016-481
• language: English
• id: 8523905
• date added to LUP: 2016-01-20 09:17:35
• date last changed: 2016-05-11 14:20:31