SHADAN WOMENS COLLEGE OF ENGINEERING AND TECHNOLOGY

TITLE:- A VANET NETWORK TRAFFIC PREDICTION MODEL USING ARTIFICIAL

INTELLIGENCE TECHNIQUES TO TAKE INTO ACCOUNT ROAD TRAFFIC PARAMETERS.
GUIDE NAME:- Dr .K. PALANI (professor , principal.)

NAME : SALEHA
BEGUM
H.NO :
21L51D5811
BRANCH :
M.TECH(CSE)
OBJECTIVE:
 The main objective of this project is to-
 Develop a network traffic prediction model for Vehicular Ad Hoc Network (VANET) that
considers road traffic parameters.
 Utilize artificial intelligence (AI) methods to learn from available data and predict the
traffic of a VANET.
 Evaluate the model in terms of accuracy and performance.
 Explore the potential of AI methods for VANET traffic prediction.
 ABSTRACT:
 Vehicular Ad hoc Networks (VANETs) enable intelligent vehicles to
communicate with each other (V2V) and with roadside units (V2R). Network
traffic prediction is an important task for optimizing the performance and
efficiency of VANETs.
 In this project build a model for predicting traffic by taking into account the
factors that can cause traffic on the roads.
 The proposed model includes the Random Forest- Gated Recurrent Unit-
Traffic Prediction algorithm (RF-GRU-NTP) to forecast traffic flow.
 This proposed model has three phases : a traffic prediction phase based on
V2R communication, a road traffic prediction phase based on V2V
communication, and a traffic prediction phase taking into account actual road
traffic based on V2V and V2R communication.
 The results show that the proposed RF-GRU-NTP model has better
performance in execution time and prediction errors than other algorithms
which used for network traffic prediction.
 INTRODUCTION

 Vehicular Ad hoc Networks (VANETs) are networks of vehicles that can communicate with each other and
with roadside units. These networks can improve traffic efficiency and road safety by providing information
and services to drivers and passengers.
 However, predicting network traffic in VANETs is challenging due to the dynamic and complex nature of
the road and network environments. Therefore, artificial intelligence methods such as machine learning and
deep learning can be useful for network traffic prediction in VANETs.
 One approach is to consider the road traffic parameters that can affect the network traffic, such as vehicle
speed, density, direction, and location.
 These parameters can be obtained from vehicle-to-vehicle (V2V) and vehicle-to-roadside (V2R)
communications. By using these parameters as features, network traffic prediction models can be trained
and tested on VANET datasets.
EXISTING SYSTEM:
 Several studies have been proposed different models to predict network traffic and
road traffic independently using learning algorithms.
 The existing system for network traffic prediction in VANETs may involve traditional
methods or simplistic approaches that do not fully utilize the potential of artificial
intelligence.
 These methods often rely on simple statistical models or rule-based algorithms that
have limitations in accurately capturing the complex relationships between road traffic
parameters and network traffic.
DISADVANTAGES:
 Less accurate data traffic flow predictions.
 Time taking process.
 Have high execution time and prediction errors.
 PROPOSED SYSTEM:
 The proposed model integrates a Random Forest- Gated Recurrent Unit- Network
Traffic Prediction algorithm (RF-GRU-NTP) to predict the network traffic flow
based on the traffic in the road and network simultaneously.
 This model has three phases including network traffic prediction based on V2R
communication, road traffic prediction based on V2V communication, and
network traffic prediction considering road traffic happening based on V2V and
V2R communication.
 ADVANTAGES:
 The RF-GRU-NTP model that has been proposed performs better in terms of execution
time
 Compared to existing algorithms used for traffic prediction, the suggested RF-GRU-NTP
model performs better in terms of prediction errors.
 LITERATURE SURVEY:
S.NO TITLE AUTHOR/YEAR METHODOLOGY CONCLUSION
1. A centralized and F. Falahatraftar, S. Propose a Naive Bayesian network congestion Obtained simulation results, applying the
dynamic network Pierre, and S. warning classification method for proposed congestion classification approach can
congestion classification Chamberland, 2021. Heterogeneous Vehicular Networks (Het improve the network’s performance in terms of
approach for VNETs) using simulated data that can be locally the packet loss ratio, average delay and average
heterogeneous vehicular applied in a fog device in a HetVNET. In throughput, especially in the dense vehicular
networks. addition, proposed a centralized and dynamic environments of Het VNET.
cloud-fog-based architecture for Het VNET. The
Naive Bayesian network congestion warning
classification method can be applied in this
architecture. Support Vector Machine (SVM), K
Nearest Neighbor (KNN) and Random Forest
classifiers, which are popular methods in
classification problems, are considered to
generate congestion warning prediction models.

2. Intelligent hybrid model T. H. H. Aldhyani, M. Propose an integrated model that combines A comparative prediction results between the
to enhance time series Alrasheedi, A. A. clustering with existing series models to proposed model and existing time series models
models for predicting Alqarni, M. Y. enhance prediction the network traffic. are presented. The comparisons indicate that the
network traffic Alzahrani, and A. M. Clustering granules are obtained using fuzzy c- presented model outperforms the opponent
Bamhdi, 2020 means to analyze the network data for models; the proposed method optimizes the deep
improving the existing time series. learning LSTM and ANFIS time series models.
The proposed methodology offers more effective
approach to the prediction of network traffic.
S.NO TITLE AUTHOR /YEAR METHODOLOGY CONCLUSION
3. Artificial W. Tong, A. Hussain, W. presents a comprehensive survey of the Finally, we present open problems and
intelligence for X. Bo, and S. Maharjan, research works that have utilized AI to research challenges that need to be addressed
vehicle-to- 2019 address various research challenges in for realizing the full potential of AI to
everything: A V2X systems advance V2X systems.
survey

4. A hybrid deep Y. Wu, H. Tan, L. Qin, B. proposes a DNN based traffic flow Experimental results demonstrated that our
learning based Ran, and Z. Jiang,2018 prediction model (DNN-BTF) to improve method outperforms the state-of-the-art
traffic flow the prediction accuracy. The DNN-BTF approaches.
prediction method model makes full use of weekly/daily
and its periodicity and spatial-temporal
understanding: characteristics of traffic flow. Inspired by
recent work in machine learning, an
attention based model was introduced
that automatically learns to determine the
importance of past traffic flow.
5. Optimized structure H.-F. Yang, T. S. Dillon, propose a novel model, stacked The evaluation results demonstrate that the
of the traffic flow and Y.-P. P. Chen,2016 autoencoder Levenberg-Marquardt proposed model with an optimized structure
forecasting model model, which is a type of deep has superior performance in traffic flow
with a deep architecture of neural network approach forecasting.
learning approach: aiming to improve forecasting accuracy.
The proposed model is designed using the
Taguchi method to develop an optimized
structure and to learn traffic flow features
through layer-by-layer feature granulation
with a greedy layer wise unsupervised
learning algorithm
 SYSTEM SPECIFICATIONS

 HARDWARE SPECIFICATIONS:  SOFTWARE SPECIFICATIONS :

• System : Pentium IV 2.4 GHz.

• Hard Disk : 40 GB. • Operating System: Windows

• Floppy Drive : 1.44 Mb.

• Programming Language: Python
• Ram : 512 Mb.
 PROBLEM STATEMENT
 One of the challenges in VANETs is to predict the network traffic flow, which is affected by both the
network conditions and the road traffic conditions.

 Existing methods for network traffic prediction either ignore the road traffic parameters or use simple
statistical models that cannot capture the complex and dynamic relationships between the network and
road traffic.

 Therefore, there is a need for a novel and intelligent network traffic prediction model that can consider
the road traffic parameters using artificial intelligence methods and provide accurate and reliable
predictions for VANETs.
SYSTEM ARCHITECTURE
ALGORITHMS
DEEP LEARNING MACHINE LEARNING

 CNN  Random Forest

 CNN+LSTM  Decision Tree
 LSTM  KNN
 Bi-LSTM  Support Vector
 RNN Machine
 GRU  Voting Classifier
MODULES:
 Data exploration: using this module, data is loaded into the system.

 Processing: Using the module, read the processing data .

 Splitting data into train & test: using this module data will be divided into train & test

 Model generation: Building the model - Deep Learning - CNN, CNN+LSTM, LSTM,
Bi-LSTM, RNN, GRU and CNN with KFoldVaildation and Machine Learning - Random
Forest, Decision Tree, KNN, Support Vector Machine and Voting Classifier.

 User signup & login: Using this module will get registration and login

 User input: Using this module will give input for prediction

 Prediction: final predicted displayed

 UML DIAGRAMS
 USECASE DIAGRAM  CLASS DIAGRAM
Splitting the data into train & test
Importing packages Exploring the dataset Data processing train
test
import() upload() process()
split()

Importing the packages

Building the model

Exploring the dataset build()

Data processing
Training the model

train()

Splitting the data into train & test

User saignup & signin

signup()
signin()
Building the model
User

Training the model

User input

choose()

User signup & signin

User input Final outcome

predict()

Final outcome
 STATE DIAGRAM
OBJECT DIAGRAM
ACTIVITY DIAGRAM SEQUENCE DIAGRAM

User System

importing the packages

Exploring the dataset

Data processing

Splitting the data into train & test

Building the model - Deep Learning - CNN, CNN+LSTM, LSTM, BiLSTM, RNN, GRU and CNN with KFoldVaildation and Machine Learning - Random Forest, Decision Tree, KNN, Support Vector Machine and Voting Classifier

Training the model

User signup & signin

User input

Prediction result
COLLABORATION DIAGRAM COMPONENT DIAGRAM

Importing Exploring Data Splitting the datase

libraries the dataset processing t into train & test

Building the
model

Training the
model

User signu
p & sgnin

User input

Prediction
 DEPLOYMENT DIAGRAM DATAFLOW DIAGRAM

User System
 FUTURE ENHANCEMENT
 However, by growing up the number of vehicles, the volume of produced data by them
would take shape of big data which in future work will implement proposed model in big
data.
 CONCLUSION
 In this project, proposed an RF-GRU-NTP model with the aim of network traffic flow
prediction based on the traffic in the road and network simultaneously.
 By combining machine learning and deep learning algorithms and analyzing various
parameters, achieved accurate predictions.
 The integration of the V2V and V2R datasets allowed us to understand the relationship
between road and network traffic, making this research the first of its kind.
 Overall, RF-GRU-NTP model has the potential to enhance traffic management systems
and contribute to the development of intelligent transportation systems.
 REFERENCES
 [1] N. Taherkhani and S. Pierre, ‘‘Improving dynamic and distributed congestion control in vehicular ad hoc networks,’’ Ad Hoc
Netw., vol. 33, pp. 112–125, Oct. 2015.
 [2] Y. Wu, H. Tan, L. Qin, B. Ran, and Z. Jiang, ‘‘A hybrid deep learning-based traffic flow prediction method and its
understanding,’’ Transp. Res. C, Emerg. Technol., vol. 90, pp. 166–180, May 2018.
 [3] H.-F. Yang, T. S. Dillon, and Y.-P. P. Chen, ‘‘Optimized structure of the traffic flow forecasting model with a deep learning
approach,’’ IEEE Trans. Neural Netw. Learn. Syst., vol. 28, no. 10, pp. 2371–2381, Oct. 2016.
 [4] W. Tong, A. Hussain, W. X. Bo, and S. Maharjan, ‘‘Artificial intelligence for vehicleto-everything: A survey,’’ IEEE Access,
vol. 7, pp. 10823–10843, 2019.
 [5] J. Chung, C. Gulcehre, K. Cho, and Y. Bengio, ‘‘Empirical evaluation of gated recurrent neural networks on sequence
modeling,’’ Dec. 2014, arXiv:1412.3555.
 [6] A. Karpathy, J. Johnson, and L. Fei-Fei, ‘‘Visualizing and understanding recurrent networks,’’ Jun. 2015, arXiv:1506.02078.
 [7] M. Coşkun, Ö. Yildirim, U. Ayşegül, and Y. Demir, ‘‘An overview of popular deep learning methods,’’ Eur. J. Techn., vol. 7,
no. 2, pp. 165–176, Dec. 2017.
 [8] K. Greff, R. K. Srivastava, J. Koutník, B. R. Steunebrink, and J. Schmidhuber, ‘‘LSTM: A search space odyssey,’’ IEEE
Trans. Neural Netw. Learn. Syst., vol. 28, no. 10, pp. 2222–2232, Oct. 2016.
 [9] R. Jozefowicz, W. Zaremba, and I. Sutskever, ‘‘An empirical exploration of recurrent network architectures,’’ in Proc. Int.
Conf. Int. Conf. Mach. Learn., Jun. 2015, pp. 2342– 2350.
 [10] P. Sun, A. Boukerche, and Y. Tao, ‘‘SSGRU: A novel hybrid stacked GRUbased traffic volume prediction approach in a
road network,’’ Comput. Commun., vol. 160, pp. 502–511, Jul. 2020
SCREENSHOTS

Fig: Home screen Fig: User Registration

Fig: User login Fig: User Input
Fig: Prediction Result
THANK YOU