Correction: Deep learning neural networks-based traffic predictors for V2X communication networks

Résumé

Confirm that the details in the "Author Contributions" section are correct. If any contributions need to be added/edited, choose the appropriate CRedIT roles from the list available here and indicate which one(s) apply. Please be aware that writing roles ("Writing -original draft" and/or "Writing -review & editing") are a requirement for authorship.Ensure all grant numbers and funding information are included and accurate (after publication it is not possible to change this information). All funders should be credited, and all grant numbers should be correctly included in this section. Note that if you add any commercial funding, please ensure that the funders involvement/non-involvement in the manuscript is declared. If you provided a positive funding statement but don't provide funding details, then the statement will be updated to say no funding was received.Confirm if the text included in the Conflict of Interest statement is correct. Please do not suggest edits to the wording of the final sentence, as this is standard for Frontiers' journal style, per our guidelines: The authors/remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.There is a discrepancy between the styling of the author names in the submission system and the manuscript. We have used [Marina Magdy Saady] instead of [Marina Magdy]. Please confirm that it is correct.The abstract should ideally be structured according to the IMRaD format (Introduction, Methods, Results and Discussion). Provide a structured abstract if possible. If your article has been copyedited by us, please provide the updated abstract based on this version.Cite the following references inside the text. " Abdellah et Reference "Sepasgozar and Pierre, " appears twice in the reference list, we have removed the duplicate one. Confirm if this is fine. confirmed yes, they are correct done confirmed yes, that is correct and I also inserted the other's authors names Abstract Vehicle-to-everything (V2X) communication is a promising technology for enhancing road safety, traffic efficiency, and the availability of infotainment services in 5G networks and beyond networks. However, the effective sharing of traffic information remains a significant challenge. To address this, AI-based systems offer potential solutions. By predicting traffic patterns on dense networks, these systems can improve traffic management, mitigate congestion, increase network safety and reliability, and improve energy efficiency. This research investigates the application of Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs) for accurate and efficient V2X traffic prediction. We explored the impact of various hyperparameters, including loss functions and optimizers, on the performance of these models. Our findings indicate that Gated Recurrent Unit (GRU) models, particularly with the Mean Squared Error (MSE) loss function and Adam optimizer, consistently outperform Long Short-Term Memory (LSTM) and Bidirectional Long Short-Term Memory (BiLSTM) models in terms of both accuracy and computational efficiency. For CNN models, the Rectified Linear Unit (ReLU) activation function, coupled with the Adam optimizer, demonstrated superior performance in terms of Root Mean Square Error (RMSE) and computational done, I cited them in the text 1- Abdellah et al., 2022b: Basel, Switzerland. 2-Ban et al., 2022: Suzhou, China 3-Fitters et al., 2021: Madrid, Spain 4-Tsourdinis et al., 2022: London, OK, that is great. Fifth-generation (5G) cellular systems and beyond are predicted to enhance quality of Q11 service (QoS), high throughput, improved network safety, increased capacity, low latency, and low cost. As the number of devices rises, so does the flow of information, making the network more difficult to manage and operate (Abdellah et al., 2022a;Sarker, 2022). For the 5G network, efficient and innovative methods are required to modify network protocols and manage resources for various services under multiple scenarios. Artificial intelligence (AI) is a leading technology in advanced intelligent technologies that allow intelligent and fast business process decisions, enhancing profitability and efficiency (Ahmed et al., 2023;Jiang et al., 2022).Recently, AI technology has been utilized in 5G wireless networks to optimize the physical layer architecture, network management, complex decision-making, and resource allocation. Big data techniques offer a great chance to grasp wireless network essentials and better comprehend 5G cellular network performance (Abubakar et al., 2020;Hassan et al., 2023). Machine learning (ML) provides or predicts new entries in most AI applications. ML solves problems like wireless network optimization and attack detection (Brik et al., 2022).Deep learning (DL) methods have proven robust for predicting network traffic and forecasting accuracy. DL algorithms based on neural networks (NNs) are promising solutions to improve prediction accuracy in data traffic flow. Many different types of NNs have been developed for various objectives; recurrent neural networks (RNNs) are made to process historical information or observations collected over specific periods; traffic patterns are an example of such observations (Sepasgozar and Pierre, 2022). A critical issue for traffic prediction is the accuracy of the forecasts to overcome the challenges of 5G mobile networks without further reducing the efficiency of the system's quality of service (QoS) (Abdellah et al., 2022a). Numerous techniques have been created to increase traffic, enhancing forecasting accuracy (Tsourdinis et al., 2022). Motivation for this study:• Traditional RNNs have problems connecting temporally distant events. • DL is a powerful method that can build accurate predictive models from vast amounts of unlabeled, unstructured data by instantly creating complicated statistical models based on their iterative output. • QoS optimization's computational difficulties.Due to these limitations, this work aims to predict Vehicleto-everything (V2X) network traffic. Healthcare, security, transportation, and medical emergencies require applications that efficiently use traffic resources. Predicting network traffic and bandwidth helps identify security and performance problems. Identifying the next steps in intensive remote patient monitoring requires a critical case to be reported to a healthcare organization within a certain timeframe. This situation results in varied information depending on the quantity and type of observations. DL techniques such as the Bidirectional Long Short-Term Memory (BiLSTM) and GRU predictors can forecast traffic volumes.The main contributions to the proposed work are as follows:• Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), and Gated Recurrent Unit (GRU) have been tested for the designed models. • A comparison was conducted between the proposed RNN models and traditional LSTM models (Abdellah et al., 2022a) regarding prediction accuracy. We evaluated prediction accuracy using root mean squared error (RMSE) and efficiency in terms of total Floating-point Operations Per Second (FLOPS) [MegaFLOPS (MFLOPS)].Several studies have explored the use of RNNs for traffic prediction, including LSTM and GRU networks (Zhou et al., 2020) applied LSTM networks for real-time traffic flow forecasting, achieving root mean square error (RMSE) values of approximately 1.23 for urban traffic data. In contrast, Kim et al. (2021) used GRU for highway traffic prediction and reported an RMSE of 0.85 under similar conditions. In Fitters et al. (2021), an architecture based on an LSTM network was investigated for predicting and focusing on irregular traffic flows. Abdellah et al. (2022b) used an LSTM network-based DL approach to forecast drone-based MEC energy consumption time series. Four cases examined accuracy as a function of learning rate. RMSE and MAPE were used to determine the best and highest average prediction accuracy. Wang et al. (2022a) suggested a spatiotemporal study of mobile network traffic and reviewed current research. Time series similarity-based graph attention networks were also proposed. Using an LSTM network with modified hyperparameters, a DL model predicts short-term traffic speeds on a parallel, multilane arterial road in an emerging country such as Vietnam (Tran, 2022). An LSTMbased practical method for accurately predicting environmental movement to improve security decision-making and path planning was first presented in Wang et al. (2022b). Then, a risk assessment was used to plan local paths. Based on these correct predictions, the risk assessment is field-based.In the domain of CNNs for traffic prediction (Li et al., 2021), used a hybrid CNN-LSTM model for traffic speed prediction, achieving an RMSE of approximately 0.91 for highway traffic, which is considered competitive for deep learning models in traffic prediction. Similarly, in Wu et al. (2020) applied a simple CNN model on urban traffic datasets and reported an RMSE around 0.98 with Rectified Linear Unit (ReLU) activation. In contrast, these studies have limitations in terms of V2X communication:• LSTM models struggle to handle massive traffic flow data simultaneously with computing and distributed storage requirements. • The current traffic forecast methods have been unsuccessful in addressing complex road segment association. To address these limitations in existing models, this study focuses on an essential issue for traffic prediction: the accuracy of forecasts to address 5G mobile network issues without reducing QoS system efficiency. Therefore, we address these design issues in this work by training alternative Deep Learning Neural Network (DLNN) architectures based on V2X packets-per-second data. These predictors don't require prior knowledge about the surrounding environmental conditions (channel statistics) and benefit from the excellent learning and generalization capabilities of DNNs. The proposed DL model will be built using LSTM, BiLSTM, and GRU, which are variations of the RNN, to solve the vanishing gradient problem. For CNNs, we will test the most powerful activation functions (ReLU, Tanh, and Sigmoid) reported in similar work. The prediction accuracy regarding root mean squared Error (RMSE) and Floating-Point Operations Per Second (FLOPS) will be assessed. The best predictors will be improving QoS demands, monitoring resource management, enhancing security, and other operational issues.Driverless and autonomous vehicles are becoming more popular because they are better for businesses and emergency services. These vehicles need constant sensor data for complex, high-speed operations and improved trajectory planning for these services. The car can use onboard sensor information for short-term trajectory decisions, but needs data from nearby vehicles for long-term decisions. Therefore, sensor data sharing is essential and requires reliable vehicle connectivity, subject to strict QoS requirements (Gao, 2022). Thus, modern wireless networks connect cars, people, infrastructure, roads, etc., via advanced communication technologies. Advanced communication technologies enable vehicle-to-everything (V2X) communication.V2X communication protocols and technologies allow vehicles to interact with roadways and users. V2X allows Vehicleto-Vehicle (V2V), Vehicle-to-Infrastructure (V2I), Vehicle-to-Pedestrian (V2P), and Vehicle-to-Cloud (V2C) interactions. Figure 1 depicts the 5G V2X communication infrastructure. v2x communication and traffic growth cause in and intelligent traffic methods are in their Thus, traffic be examined to test a new traffic on (Gao, overcome the issues with 5G mobile networks and further of the quality of service (QoS) solutions are required to improve the of traffic prediction. DL algorithms have effective in predicting road traffic to statistical techniques and Deep learning including neural networks Recurrent Neural Networks and are utilized to predict various of traffic, such as congestion, and These models are of learning complex, relationships in making them for traffic prediction, which a of environmental and 1 a comparison between RNNs and CNNs for traffic prediction in V2X studies used deep neural network to predict based on historical including et al., LSTM et al., and These studies predicted were also by for decision-making and the of to process the of predictors in study proposed and with and tested with RNN and CNN models for V2X traffic prediction on V2X All and have been conducted using the V2X system was to the DL training Then, the training the collected V2X was and to the DL model for prediction. The DL model used for the training and for the data by and values to The network the training and a loss function, according to the to the error between and predicted the training models, the gradient of the loss function was and the network and were This was the error was as as possible. The test network requires test to the model in the following Using used to from data in traffic prediction. to handle and data. is a and system with traffic flow data from different as traffic and vehicle patterns and in the of learning patterns from traffic data RNNs and can forecast traffic V2X systems with or data from CNNs can or to traffic congestion, or models can predict or based on the historical of in a or to the conditions of or such as or to predict the time for a for current and historical traffic the first RNN using traditional optimization techniques a study between the proposed and GRU models and the traditional LSTM model (Abdellah et al., 2022a) performance using RMSE and total that the proposed models provide results for the Figure the for the proposed work collected a DL model training from the V2X system (Abdellah et al., 2022a). All and have been conducted in Figure the V2X The model is a V2X system for a the of the on the is The model helps a path The training is QoS is to be a critical and issue in networks. Therefore, of the essential QoS is throughput, which we collected and used as the for training the ML model with the flow that In this we have made for V2X traffic, and the is used as to the ML training be in a specific for a forecasting the data quality and the model used for data was used to data and manage The data were then to ensure all were within the The ML training the network data that has been The network is using such as the network architecture, training and a error between the predicted and The is data are used to test the network, and the results are then to the The data are the network has an the predicted is to determine the this we to improve the learning process to a robust Therefore, we use various using different loss functions in the layer that will be in various depending on the number of per and in a V2X This will be done by traditional loss such as with more robust loss functions that will in robust learning networks with better in the of different of The prediction accuracy regarding RMSE and total will be The proposed is to existing models using these traffic flow forecast The error will be used to model performance in The best will be used for improving the QoS demands, monitoring resource management, enhancing security, and other operational The in network learning and we use the traditional loss function, mean squared error The of in statistical models is by The average squared between the and values is The is to in a model that has no The of the model with the The mean squared which is for the mean squared is to the To determine the the the predicted and square that that for all observations. Then, all of squared values and by the number of observations. The is the of the squared which the by the The for is the is the is the predicted the number of function has with no local and it the model for making by In contrast, the are not as the error will be and it is by we built a layer that other loss functions such as error of squared to the best network Our model was built using mean error as a loss is a popular with root mean squared error the error the predicted the in are and and RMSE or the mean error to the square of the error In different are not more or but the increase with the increase in is as the average of the error The function is a function that a number Therefore, the between an and a predicted can be positive or and will be positive the The can be as 1 of is that the are better as it does not the model by the error In contrast, are that it is the and be a local the model was built using the of squared or of or is the between the and predicted it performance according to the of squared The the following is the between the of the and the predicted aims to the the the better the this we study the of the hyperparameters, which loss functions and and and on different RNN models BiLSTM, and for V2X traffic prediction. The performance is evaluated in terms of accuracy using as in and with the system in Floating-Point Operations Per Second (FLOPS) as this we study the performance of models for V2X traffic prediction with loss functions and and and on activation functions Linear Unit Tanh, and Tanh, and The performance is also using which is in and total Floating-point Operations Per Second (FLOPS) as in to the results in for our RNN models, for the Adam with loss function, the results indicate that for GRU best with the RMSE of and a computational of As the GRU consistently RMSE values to LSTM and BiLSTM, making it the most efficient model in terms of both prediction accuracy and computational the loss function, the performance is similar to the GRU the best between accuracy and efficiency. loss function results also that GRU by the but the has more the GRU For the optimizer, the with the Adam they are by RMSE values and increased computational For the loss function, the model the best with an RMSE of it requires more computational resources the the performance is the GRU best to LSTM and For the optimizer, results that with Adam and RMSE values all models. In the results are not as with other loss in contrast, with better RMSE with and In the results from RNN indicate that the GRU model both LSTM and models in terms of prediction accuracy (RMSE) and efficiency This is particularly with as the loss function, by it is essential to that the of the a with Adam the best performance the different loss results for the CNN models the the Adam with the loss function, the activation function provides the best between prediction accuracy and system all with the RMSE of and All activation functions with the loss function, with variations between the loss function, the performance is with achieving the RMSE values and a between accuracy and performance with and is similar to the Adam optimizer, but the loss function However, requires computational resources For a data the and then activation functions the best accuracy with the activation The loss function that the activation function accuracy the and activation but it requires more computational resources. the loss function, a between accuracy and efficiency, with by The performance of CNN models using the is to that of Adam and For the loss the activation function the best performance in terms of accuracy. In contrast, the computational is which the between accuracy and computational efficiency with for CNN we can that other activation functions and Sigmoid) different loss functions and in terms of both RMSE and computational efficiency. the best it with high computational resources. The activation function a for accuracy and efficiency. The Adam results in the best with in terms of but with the most efficient model in terms of both accuracy and computational the of RMSE and according to the for the Recurrent Neural Network models, and the for Convolutional Neural Network models as of RNN and CNN models in the of V2X traffic prediction that model type has depending on the use The RNN models, particularly the Gated Recurrent Unit outperform CNNs in terms of prediction accuracy RMSE but in contrast, they need more computational resources In contrast, CNN models computational Rectified Linear Unit (ReLU) activation functions and the Adam optimizer, which them a more efficient for applications computational is a critical CNNs not consistently the low RMSE values that RNNs can they offer a between accuracy and computational efficiency. the to choose between RNN and CNN models on the specific requirements of the traffic prediction such as the required computational and the comparison of the proposed RNN and CNN models with a critical of deep learning models in practical to V2X communication are to various including from other and sensor et al., These and the data which can the performance of prediction models that were on or of in the data can have on traffic prediction • The most impact is an increase in prediction the patterns and relationships that models like RNNs and CNNs are designed to This can to a in the reliable for critical like or traffic Deep neural networks have a high to complex which can them to the in the training data as if it were a model in this will as the will from in the training • A model that has not been to conditions training to to different operational urban and the quality and ensure the practical of our proposed it is to their A proposed architecture, particularly the RNN BiLSTM, that can to The GRU and LSTM are designed to handle long-term and can to or focusing on the more in the traffic data. The allow these models to the flow of information, By of data RNNs a of A data within a has a impact on the prediction, as the is based on the of the CNNs are powerful for their to can be that of RNNs they are or with as they or local further enhance can be which also a for our of the training This as a powerful the model to that are to and to data (Sepasgozar and Pierre, 2022). • As explored in this certain loss functions like are more robust to such loss functions in could to more and average on the V2X data it is the prediction the results presented in the superior performance of our GRU and CNN models under data we that to is a requirement for V2X The of particularly suggest a to A of this following the will be a of our research to these models from a to a practical for intelligent a the models, RNNs BiLSTM, and and CNNs, for accurate and efficient V2X traffic prediction. We explored the impact of various hyperparameters, including loss functions and and and on the performance of these models. The proposed traffic data to predict traffic patterns to improve forecasting and decision-making in V2X networks. The prediction problems are in different cases depending on the number of per The prediction accuracy is in terms of RMSE and the number of A critical of this study is the of the under conditions. By the models data with which sensor and to V2X the following was the increased the prediction error (RMSE) all models, on data However, the recurrent particularly the GRU and BiLSTM, demonstrated superior a in accuracy to the CNN models as the This to to the and of the use of these as a for traffic systems that on V2X data the GRU model is the for V2X traffic prediction, the best between high low computational and essential data For the the Adam consistently and in terms of both accuracy and efficiency. For the loss function is a can be in specific is

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0