Hybrid and Advanced Technologies – S.
Prasad Jones Christydass, Nurhayati Nurhayati and
S. Kannadhasan. (Eds) © 2025 the Authors, ISBN 978-1-032-90669-0
Open Access: www.taylorfrancis.com, CC BY-NC-ND 4.0 license
Development of communication protocol in 6G network with
quality of service and dynamic resource allocation through seamless
connectivity
Sumit Kumar Kapoor
Department of Computer Science and Engineering, Poornima University, Jaipur, India
S. Sophia
Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences,
Coimbatore, Tamil Nadu, India
Ch. Venkata Krishna Reddy
Department of Electrical and Electronics Engineering, Chaitanya Bharathi Institute of Technology,
Hyderabad, Telangana, India
M. Rajendiran
Department of Computer Science and Engineering, Panimalar Engineering College, Poonamallee,
Chennai, Tamil Nadu, India
T.R. Vijaya Lakshmi
Department of Electronics and Communication Engineering, Mahatma Gandhi Institute of Technology,
Hyderabad, Telangana, India
Ramya Maranan
Department of Research and Innovation, Saveetha School of Engineering, SIMATS, Chennai, Tamil
Nadu, India
ABSTRACT: Quick urbanization and change in the smart city facilities owing to tech-
nology call for new and innovative means of pollution monitoring and managing. In this
regard, the current study proposes a communication protocol in the 6 G networks combined
with dynamic resource allocation for pollution monitoring. Additionally, the study adopts
machine learning technologies including ANN , DT , LR , and SVM to accurately predict
pollution levels and trends in smart city situations. The solution allows for the application of
an accurate machine-learning-based approach to pollution prediction while offering a
communication protocol that can ensure interruption-free connection between the pollution
sensor networks and the central pollution monitoring system. The proposed resource allo-
cation scheme guarantees that the network resources including, but not restricted to, com-
puting power and bandwidth are used efficiently. The performance of each of the four
models in accurately predicting pollution is tested by calculating precision, recall, F1 score,
and accuracy metrics. The performance of each of the models in distinguishing between
classes is analyzed closely by drawing the confusion matrix. The results show that all models
have a superior level of accuracy with ANN being slightly more accurate than the other
models. The results suggest that using advanced communication protocols combined with
accurate machine learning technologies can help with pollution monitoring. In this way, the
relevant stakeholders can make informed decisions prior to the onset of adverse pollution
levels.
186 DOI: 10.1201/9781003559139-24
Keywords: smart cities, pollution monitoring, communication protocols, machine learning,
environmental sustainability
1 INTRODUCTION
In recent decades, urbanization and technological advancements have transformed cities into
innovative, interconnected environments, yet they have also led to increased pollution and
environmental degradation. As urban populations continue to grow, the challenges asso-
ciated with pollution are becoming more complex. In response, the concept of smart cities
has emerged, leveraging technology to enhance environmental performance and quality of
life [1–3]. This paper explores the promising 6G communication technology and a dedicated
communication protocol for monitoring pollution in smart cities, along with machine
learning applications for pollution prediction. Smart cities integrate IoT devices and sensors
to monitor environmental parameters like air quality and pollution levels, generating valu-
able data for decision-making and policy formation. However, existing communication
protocols such as Wi-Fi and cellular networks face limitations in dense urban areas,
including bandwidth issues and reliability concerns[4–6]. The upcoming 6G networks pro-
mise to overcome these challenges with high-speed connectivity and low latency, com-
plemented by machine learning techniques that facilitate real-time pollution analysis and
proactive management. This research addresses the need for an optimized communication
protocol for 6G networks, ensuring efficient and reliable transmission of pollution data in
smart cities. It aims to develop a protocol architecture that meets quality of service
requirements and dynamic resource allocation for pollution monitoring applications. The
research evaluates the protocol’s performance under varying network conditions using
simulated data, aiming to improve environmental and public health outcomes in smart cities
[7–9].
2 METHODOLOGY
Figure 1. Proposed methodology.
From Figure 1, The current project methodology involves deploying pollution sensors,
establishing a 6G communication infrastructure, developing a communication protocol, and
utilizing machine learning models for pollution prediction and alert generation. These sen-
sors are installed across the city to measure air quality metrics such as PM2.5, PM10, NO2,
SO2, CO, and O3, providing real-time data on particulate matter and other pollutants. The
6G network facilitates rapid communication, ensuring sensors remain connected to the
central monitoring system. This system supports dynamic resource allocation and quality of
187
service tools for prioritizing pollution data transmission. Machine learning models, including
Artificial Neural Networks and Support Vector Machines, analyze data from sensors, traf-
fic, time, and sensor locations to predict pollution levels. Alerts are generated if pollution
exceeds safe levels. A central monitoring system collects and analyzes real-time pollution
data, identifying hotspots and trends. It uses advanced data analysis tools to inform city
authorities and environmental agencies promptly. Data preprocessing ensures the quality
and reliability of machine learning models. This includes cleaning, normalization, feature
extraction, and data partitioning to optimize pollution monitoring in smart cities.
3 RESULT AND DISCUSSION
The testing results shown in Figure 2 highlights that the selected ML models are very effi-
cient in predicting the pollution levels for area pollution degree level, where several imple-
mentations programmed in various algorithms showcased remarkable accuracy. First, the
ANN plot shows the high-predictions level, reaching 98.56% of accuracy. DT, in its turn,
gained a prediction accuracy of 94.5%, which is attributed to its close ties with determining
the pollution level of the whole area and understanding the factors that cannot be accounted
for in pollution control. LR and SVM also present a decent plot prediction, with 91.2% of
accuracy and 96.23%, respectively. Evaluation of the outcomes may indicate that all models
present high prediction accuracies, confirming the robustness of the models.
Figure 2. Accuracy of each model.
Overall, the machine learning models chosen for implementation in the study showed high
levels of accuracy, as evidenced by all proposed models being able to predict the pollution
environment. Therefore, the implemented pollution prediction may be used for a further
prediction derived from the previous data employing machine learning tools and supporting
the implementation of decision-making measures.
According to our evaluation as shown in Figure 3, artificial neural network, or ANN
model demonstrated perfect results exceeding all other models and reached a precision of
97.5%, recall of 96.5%, and F1 score of 97.0% with overall accuracy of 98.56%. The results
of the decision tree model, or DT, show that this model has also demonstrated
acceptable performance. Specifically, the decision tree reached a precision of 94.5%, recall of
93.0%, and F1 score of 93.7% with overall accuracy of 94.4%. Logistic regression, or LR,
188
and support vector machine, or SVM, have also shown commendable performance though
slightly less effective compared to ANN and DT. Thus, the logistic regression reached a
precision of 91.2%, recall of 90.5%, a F1 score of 90.8%, and an accuracy of 91.2%. In
particular, the support vector machine reached a precision of 96.2%, recall of 95.5%, and
F1 score of 95.8% with 96.23% accuracy. In summary, all of these metrics demonstrate that
each of the ML model is reliable and effective in predicting pollution levels, which can be
used to inform officials and the public on this problem for more proactive management and
public health protection.
Figure 3. Performance score of each model.
Figure 4. Confusion matrices of each model.
Confusion matrices help to provide a full image of the performance of each machine
learning model in the prediction of pollution outcomes as shown in Figure 4. In terms of
pollution monitoring, such a matrix compares the actual levels of pollution with the
189
estimations made by different ML tools, dividing the instances into true and false ones
in both directions. For example, according to the confusion matrix of the Artificial
Neural Network , 975 instances were true negatives or correctly classified as negative,
while 25 were false positives or erroneously classified as positive. Also, 965 was the
number of true positives or correctly classified as positive, and 35 was the number of
false negatives or erroneously classified as negative . This information allows for a more
specific evaluation of the model in terms of precision, recall, and accuracy. By viewing
the confusion matrices made for all models, the stakeholders can see where the model
performs well and poorly and make decisions or implement measures to improve pol-
lution prediction. As a result, confusion matrices are effective tools in understanding the
mechanisms behind ML performance in the prediction of environmental outcomes.
They are valuable because their deployment makes it possible to assess the reliability
and effectiveness of ML tools when working on the prediction of pollution as a smart
city characteristic.
4 CONCLUSION
The development, and implementation of a communication protocol in 6G networks,
along with dynamic resource allocation, have great implications for the monitoring of
pollution within a smart city setting. Through the incorporation of machine learning
models, particularly ANNs, DT, LR, and SVM, attempted to demonstrate that the
technology can produce sufficiently accurate prediction and successfully identify pollu-
tion trends . The viability and efficiency of the developed model have been proven
through the calculations of the values of precision, recall, F1 score, and accuracy
achieved . Additionally, the capacity to allocate dynamic resources proposed within the
communication protocol ensures that the available resources are used efficiently with
high throughput, reduced latency, and an improved quality of service. The use of the
confusion matrices also adds much to the assessment of each of the models that were
reviewed, therefore, providing valuable information for stakeholders, including health-
care professionals, and administrators, to consider when managing pollution sources. In
the end, the use of innovative approaches such as 6G communication networks and
machine learning is essential to addressing the issue of pollution that modern smart cities
are facing. Using advanced technologies, cities can develop effectively and become
competitive while preserving the well-being of human populations, as well as the envir-
onment, making them sustainable for future generations.
REFERENCES
[1] Rahman, A., Hasan, K., Kundu, D., Islam, M. J., Debnath, T., Band, S. S., and Kumar, N. (2023). On
the ICN-IoT with federated learning integration of communication: Concepts, security-privacy issues,
applications, and future perspectives. Future Generation Computer Systems, 138, 61–88. https://doi.org/
10.1016/j.future.2022.08.004
[2] Taimoor, S., Ferdouse, L., and Ejaz, W. (2022). Holistic resource management in UAV-assisted wire-
less networks: An optimization perspective. Journal of Network and Computer Applications, 205(May),
103439. https://doi.org/10.1016/j.jnca.2022.103439
[3] Sur, S. N., and Bera, R. (2021). Intelligent reflecting surface assisted MIMO communication system: A
review. Physical Communication, 47, 101386. https://doi.org/10.1016/j.phycom.2021.101386
[4] Sadiki, A., Bentahar, J., Dssouli, R., En-Nouaary, A., and Otrok, H. (2023). Deep reinforcement
learning for the computation offloading in MIMO-based Edge Computing. Ad Hoc Networks, 141
(August 2022), 103080. https://doi.org/10.1016/j.adhoc.2022.103080
190
[5] Ali, A., Ali, A., Hussain Zahid, A., Mustafa, G., and Waseem Razzaq, M. (2022). A non linear model
to analyze the DC performance of SiC MESFET. Ain Shams Engineering Journal, 13(1), 101506.
https://doi.org/10.1016/j.asej.2021.05.020
[6] Ahmad, M., and Shin, S. Y. (2023). Wavelet-based massive MIMO-NOMA with advanced channel
estimation and detection powered by deep learning. Physical Communication, 61(March), 102189.
https://doi.org/10.1016/j.phycom.2023.102189
[7] van Hilten, M., and Wolfert, S. (2022). 5G in agri-food - A review on current status, opportunities and
challenges. Computers and Electronics in Agriculture, 201(August), 107291. https://doi.org/10.1016/j.
compag.2022.107291
[8] Sharma, H., and Kumar, N. (2023). Deep learning based physical layer security for terrestrial com-
munications in 5G and beyond networks: A survey. Physical Communication, 57, 102002. https://doi.
org/10.1016/j.phycom.2023.102002
[9] Khan, H. U., Sohail, M., Ali, F., Nazir, S., Ghadi, Y. Y., and Ullah, I. (2023). Prioritizing the multi-
criterial features based on comparative approaches for enhancing security of IoT devices. Physical
Communication, 59, 102084. https://doi.org/10.1016/j.phycom.2023.102084
191
