Modeling Physical Channel For Vehicle-to-vehicle Communications And Network Optimization

Vehicle-to-vehicle (V2V) communication systems have received ever-increasingattention since IEEE802.11p is proposed and approved. Understanding physical channel andoptimizing network performance are two key issues that needed to be solved before V2Vcommunication system comes into use.As for physical channel, the scenarios in V2V communication are quite different fromthat in traditional wireless communications, e.g. short-lived link holding time, high-speedmotion of transmitters and receivers, and low height of antennas. Due to these significantdifferences, V2V wireless channels are characterized by distinctive spatial as well as temporalimpulse responses. Plenty of work has been done on V2V channel and the method arecategorized into modeling and measurement. Modeling describe objects in V2Vcommunication scenario by formulas and calculate results by program while measurementrequires kinds of devices. Modeling costs less money and time than measurement whilemeasure provide more accurate data. The modeling can be classified into geometry-baseddeterministic model (GBDM) and stochastic model while the latter can be further divided intogeometry-based stochastic model (GBSM) and non-geometrical stochastic model (NGSM).The GBDMs can provide the most accurate data among these models. However, the GBDMsneed massive data about communication scenario and are most time-consuming.Fromperspective of geometrical optics theory, the paper proposes to comprehensively model V2Vphysical channel and integrate the proposed models with CarSim and Simulink in order toobtain necessary scenario parameters in real time. A series of traffic and environmentalmodels, including vehicle, road, building, tree, and weather are presented in this paper.Moreover, image method is used to deter-mine potential wave propagation paths consistingof line-of-sight, reflections, diffractions, scatterings and their combinations. Furthermore,typical and prime channel characteristics such as impulse response and average received power are analyzed theoretically, which drive the subsequent V2V communicationsimulations under various combinations of traffic situation and network scenario.As for network optimization, we can evaluate network performance from different typesof standards, e.g. throughput. As for the issue in VANET, some researchers has proposed todynamically adapt contention window size and transmission power based on vehicle densityaiming to improve throughput and decrease delay. By using Partially Observed MarkovGames (POMG), which is derived from Partially Observed Markov Decision Process(POMDP), we proposed a new scheme for dynamic adaptation of transmission range,contention window size and bit rate to enhance network performance, e.g. throughput,utilization, delay and number of nodes, based on traffic density. It is a multi-objectiveoptimization. The experiment showed that these objectives are interacted. When one objectiveis enhanced, other objectives are weakened. The network performance demands varies sincethe applications are different. For example, safety application need very short network delayand visual telephone application requires high-throughput network to supply massive imagedata transmission. The proposed model can be applied to kinds of VANET communicationscenarios. Extensive simulations have demonstrated that our scheme significantly improvenetwork performance compared with the other schemes.