802.11p Optimization for Delay Sensitive in Non-Safety Messages in VANETs

ABSTRACT

In the highway scenario, vehicles travel in a line where infrequent RSUs are high while relative speed between vehicles is low. Highway traffic can be evaluated for vehicles moving in the same direction and also in the opposite direction. In contrast, in the city driving scenario vehicles travel at low speed and have a frequent stop-start. There are more numbers of RSU in the city compared to the unpredictable highway scenario. Then, routing in the city is more complex because of all the large obstacles such as buildings and the behavior of car drivers is unpredictable.
Vehicles travel in a straight line in highway scenario which number of RSUs is high while velocity between vehicles is low. Highway traffic can be measured for both vehicles moving in a similar and opposite directions. In contrast, the city drivers travel at low speed and have a continous stop and start. The numbers of RSU are more in the city contrasted with highway environments. Subsequently, routing in city environment is more complicated due to many buildings and the unpredicted driving movements.
Traffic optimization is one approach to keep up the current congestion control protocols and other traffic parameters for formulating new enhanced congestion control protocols [9]. The data transmission between vehicles and RSUs must be reliable in triggering changes in VANET driving environments. A tool called Multi-Objective Network Optimization and Analysis Tool (MONOPATI) created by [10] utilizing Snealing Algorithm (SA) is as an instrument in optimizing the network in spite the absence of any network topology changes. It does not depend entirely on irregularities exhibit on any changes to the parameters to enhance the mobility execution. Moreover, the system need to be intelligently improved the zero error environments. Consequently, this paper shows an optimizatin of VANET in delay sensitive applications for both city and highway traffic scenarios using Taguchi method.

RESEARCH METHOD
In 1960, Dr. Genichi Taguchi intitiated the Taguchi method [11] focusing on low cost of producing high quality product while reducing the production process applying a robust experimental design [12]. The Taguchi method is an experimental design to evaluate various factors influencing the characteristics in production process operations. Taguchi's method is an optimization procedure based on orthogonal array (OA) in getting near-optimal settings [13]. OA which is an imperative parameter in Taguchi's method discloses systematically identify the control parameters of the experimental run. Taguchi's method is generally connected to manufacturing production processes and other engineering fields, such as electromagnetic, power electronics and wireless communications [14]. The Taguchi method contains three phases which are clarified in Figure 1

Basic Design Specification
Design of experiment (DOE) in Taguchi method is connected to congestion control in VANETs. This method reduces the variation in congestion control processes. In Figure 2, Taguchi classified variables into two, first the control factors that can be controlled while noise factors which are hard to control. The reason for the Taguchi experimental design is to have the validating factors set at maximum levels and compared with the existing execution. Finally it can be round up as a precise verification experiment with factors set at the maximum levels should be achieved to validate the earlier results.

Methodology over Congestion Control
OMNet++ is an open source software that enables the design of the proposed congestion control framework of traffic system network to the users' viewpoint. Some of the traffic application formats for the traffic generator are UDP, TCP, ViDeo, etc. The selected reference traffic type for this experiment is UDP. The simulation scenario was run over the 250 simulated seconds in a wireless vehicular mobility environment. Table I

Experiments Phase
This experiment concerns maximizing the control factors in VANET congestion controls in accomplishing least end-to-end delay. Next, the control factors that should be maximized are packet size and distance between RSUs. The three noise factors are number of cars, interval time of packets generation and cars' mobility speed. Levels of variations of control factors are presented in Table II.
Packet sizes are from 25KB up to 125KB. Five levels of RSU distance both for urban and highway scenarios are expressed in Table IV. 802.11p was applied for MAC protocols whereas AODV was identified for routing protocols. Levels of variations of noise factors are displayed in Table 2. The experimental table for control factors is based on orthogonal array L8 Taguchi configuration which is designed for four factors and each factor has three levels. While for noise factors, orthogonal array L4 Taguchi configuration is used which has three factors and each factor has two levels.
Experiment attempts are done utilizing OMNeT++ ver 4.6 simulator [16] running under UBUNTU 14.04.2 LTS operating system. All medium access control (MAC) and routing protocols are based on INET framework [17], [18] and INET-MANET [19] of the OMNeT++ [23]- [24]. All control factors and noise factors are presented in Table 2, Table 3 and Table 4 as the experiments parameter sets. Simulation time of each experiment is set at 250 seconds and 3 (RNG) random network/seed generation is performed [15].   B  C  D  1  1  1  1  1  2  1  2  2  2  3  1  3  3  3  4  2  1  2  3  5  2  2  3  1  6  2  3  1  2  7  3  1  3  2  8  3  2  1  3 To determine the impact each factor has on the output, the signal-to-noise (SN) ratio should be figured out for each experiment conducted. The SN value represents the magnitude of the average procedure constrasted with its variation. There are three sorts of SN ratio which can be computed depending on the various types of execution attributes. For maximizing the execution attributes, the following SN ratio, which is called the larger-the-better is utilized. In order to minimize the execution of the system, the following SN ratio, which is called smaller-the-better, should be calculated utilizing Equation 1, 2 and 3.
Where y = mean response of the experiment, i = experiment number, u = trial number, Ni = number of trials for experiment i. The third case is for nominal-the-best situation when a specified value is most desired. The SN ratio should be calculated as follow: For this experiment, the SN ratio smaller-the-better is utilized for end-to-end delay evaluation, as shown in Table 5. The experiment is aimed for non-safety applications in VANETs. Essentially there is an interest on non-safety applications for low quality of service (QoS) which are delay sensitive. Therefore, for maximum execution, the larger-the-better execution metric for delay must be considered for non-safety applications. The delay sensitive should be evaluated as execution metrics to end-to-end delay.

RESULTS AND ANALYSIS
This section will discuss the outcomes from the optimization configuration and simulation in OMNeT++. The end-to-end delay execution attributes is illustrated in Figure 6 utilizing mean SN Plot for end-to-end delay response versus control factors.
Based on Figures 6 and 7, 15.6% on average reduced the delay sensitivity applications after the optimized the traffic congestion control. There is a possibility that the vehicles network connection suffers from latency at point 25KB to 50KB while increased in vehicles speed. ON the top of that, AODV multi-hop data transmission of V2I can lead to extra delay as a packet may need to wait for prior missing packets during data re-transmission. But, this did not reduce the efficiency on traffic capacity usage. The plotted graphs also concluded the S/N smaller is better, applying Taguchi method in order to optimize and reduce end-to-end delay. Table 6. Simulation Results of Matrix VANETs Experiments From the experiment, the optimization of congestion control in VANET in minimizing end-to-end delay is high from point of 25KB to 50KB packet size. The packet size has a big impact on optimizing the congestion control performance with 28.37% on average in term of optimization percentage. The effect of parameters in the experiment on the end-to-end delay performance matrix can be seen through the variation percentage of SN ratio values in Table 6. It exhibites the direct and indirect impact on VANET networks performance attributes. A proposed optimization technique is relevant for a various design and control factors such as MAC and routing protocols and VANET topology and environments. Taguchi optimization technique has been proposed in is paper. For future research, more parameters and variables can be associated with congestion control optimization experiments.
It is concluded that seamless performance can be achieved with delay sensitive application in vehicular ad hoc wireless access network of mobile user traffic aiming to maximize the overall effectiveness of non-safety packet transmission and improved network quality services in terms of end-to-end delay sensitivity. In this paper, a typical framework that has been setup examined the performance differences of AODV routing deployment only under two different scenarios that is before and after optimization of AODV routing in urban or city driving conditions. Our simulation results show the AODV routing deployment in the urban or city driving scenario has pros and cons in terms of the end-to-end delay transmission received. However, one positive effect is, the congestion control of the particular framework is improved drastically after optimization for the vehicular ad hoc network of user or application level.
As a conclusion the traffic bandwidth utilization could be materialized in delay sensitive application in VANET environments. These could be optimized the efficiency of non safety data transmission and QoS towards end-to-end delay sensitivity. In this paper, the setup experiments examined the AODV routing performance before and after the proposed optimization in city driving environments. The simulation results shown the impacts in AODV routing deployment in terms of end-to-end delay packet transmission received. Nevertheless, the bandwidth ulitization on the network congestion is increased significantly after optimization techniques done for VANET at network application level.