UNIVERSITY OF GUJRAT

DEPARTMENT OF IT (EVENING), HAFIZ HAYAT

CAMPUS

Proposal
AI-Driven Security for Fog and Edge Networks

Course: System and Network Administration

Course Code: IT 381
Program: BSIT (Evening)

Submitted by:
Asfand Ahmed Baig
Roll No: 21014156-040
Submitted To:
SIR JUNAID MUZAFFAR

Research Proposal: AI-Driven Security for Fog and Edge Networks

Page 1 of 16
Description of Research Work

Fog and edge computing are transforming the landscape of IT by bringing computation and data
storage closer to the source of data. This shift addresses the limitations of centralized cloud
computing, such as latency, bandwidth constraints, and privacy concerns. Fog computing
extends cloud services to the edge of the network, while edge computing processes data on local
devices or servers. The convergence of Artificial Intelligence (AI) with these paradigms has the
potential to enhance security measures, making networks more resilient to threats and more
efficient in data handling (Zhang & Qi, 2022).

The concept of fog computing was first introduced by Cisco in 2012 to address the needs of IoT
applications requiring low latency and real-time processing (Bonomi et al., 2012). Since then, the
integration of AI with fog and edge networks has emerged as a pivotal area of research, focusing
on improving security protocols and data management practices. AI-driven security mechanisms,
such as machine learning algorithms for threat detection and anomaly analysis, promise to
enhance the robustness and reliability of fog and edge networks (Liang & Shi, 2022).

Overview of Fog and Edge Networks: Fog and edge computing represent a paradigm shift in
how data is processed and managed in IoT ecosystems. Unlike traditional cloud computing,
which centralizes data processing in remote servers, fog and edge computing bring
computational capabilities closer to where data is generated. Edge nodes, situated near IoT
devices, and fog devices, located at the edge of the network infrastructure, facilitate quicker data
processing and reduced latency. This proximity not only improves the performance of latency-
sensitive applications but also optimizes bandwidth usage by processing data locally before
transmitting it to centralized cloud servers.

Importance of Security in IoT and Edge Computing: The rapid proliferation of IoT devices
and the widespread adoption of edge computing have introduced new challenges regarding data
security. IoT devices, ranging from sensors in smart homes to industrial machinery in smart
factories, collect and transmit vast amounts of sensitive data. These devices often operate in
diverse and distributed environments, increasing their susceptibility to cybersecurity threats such
as unauthorized access, data breaches, malware attacks, and denial-of-service (DoS) incidents.
Securing these devices and the data they generate is crucial to maintaining the integrity,
confidentiality, and availability of IoT systems.

Role of AI in Enhancing Security: Artificial Intelligence (AI) has emerged as a powerful tool
in fortifying the security posture of fog and edge networks. AI techniques, particularly machine
learning algorithms, enable these networks to autonomously detect, analyze, and respond to
security incidents in real-time. By leveraging AI, fog and edge devices can continuously monitor
data patterns, identify anomalies that indicate potential threats, and implement proactive security
measures before breaches occur. Moreover, AI-driven security solutions can adapt dynamically
to evolving cyber threats, enhancing the resilience of IoT environments against sophisticated
attacks.

AI algorithms such as deep learning, reinforcement learning, and anomaly detection play pivotal
roles in augmenting security measures. Deep learning models, for instance, can analyze complex

Page 2 of 16
datasets to recognize patterns associated with normal and abnormal behavior within IoT
networks. Reinforcement learning algorithms enable edge devices to learn optimal security
policies through interactions with their environments, improving decision-making capabilities in
response to emerging threats. Anomaly detection techniques empower fog nodes to detect
deviations from expected data patterns, triggering immediate responses to mitigate potential
security risks.

In essence, AI not only enhances the efficiency and effectiveness of security protocols in fog and
edge networks but also empowers IoT ecosystems to achieve higher levels of resilience and
robustness against diverse cyber threats. By integrating AI-driven security solutions,
organizations can safeguard their IoT deployments while maintaining operational continuity and
data integrity in an increasingly interconnected digital landscape.

Primary Objectives

Objective 1: Evaluation of AI Techniques

 Description: This objective focuses on assessing the effectiveness of AI techniques in

enhancing security for fog and edge networks. AI techniques may include machine
learning algorithms (e.g., supervised learning for classification, unsupervised learning for
anomaly detection), deep learning models (e.g., convolutional neural networks for image-
based security, recurrent neural networks for time-series data analysis), and natural
language processing for analyzing security logs and communications.
 Methodology: Conduct a comprehensive literature review to identify state-of-the-art AI
techniques applicable to network security in fog and edge computing environments.
Develop criteria for evaluating these techniques based on metrics such as detection
accuracy, false positive rates, scalability, and resource efficiency.
 Expected Outcome: By the end of this objective, you should have a clear understanding
of which AI techniques are most suitable for addressing specific security challenges in
fog and edge networks.

· Objective 2: Development of AI Models

 Description: This objective involves the development and customization of AI models

specifically designed to address security threats in fog and edge networks. AI models
may include supervised models trained on labeled datasets of network traffic,
unsupervised models for anomaly detection in IoT device behavior, or reinforcement
learning approaches for adaptive security policy optimization.
 Methodology: Design experiments or simulations to train and validate AI models using
representative datasets from fog and edge network environments. Implement algorithms
and frameworks suitable for deployment on resource-constrained edge devices while
ensuring robustness and real-time responsiveness.

Page 3 of 16
  Expected Outcome: Successful development of AI models tailored to detect and
mitigate security threats in fog and edge networks, demonstrating improvements over
traditional security approaches.

· Objective 3: Performance Evaluation

 Description: This objective aims to evaluate the performance impact of AI-driven

security mechanisms on the operational characteristics of fog and edge networks.
Performance metrics include latency (response time), throughput (data transfer rate), and
overall system reliability under varying network loads and attack scenarios.
 Methodology: Conduct experiments using testbeds or simulations to measure the
performance of AI-driven security mechanisms in realistic network environments.
Compare performance metrics with baseline measurements obtained from non-AI-driven
or traditional security solutions.
 Expected Outcome: Quantitative assessment of how AI-driven security solutions
influence network performance, providing insights into their feasibility and practical
implications for deployment in fog and edge computing architectures.

· Objective 4: Comparative Analysis

 Description: This objective involves comparing AI-driven security solutions with

traditional methods to identify strengths, weaknesses, and potential areas for
improvement. Traditional methods may include rule-based systems, signature-based
detection, and heuristic approaches commonly used in network security.
 Methodology: Conduct a systematic review and empirical evaluation to compare the
effectiveness, scalability, and adaptability of AI-driven security solutions against
traditional methods. Use benchmark datasets and standardized evaluation metrics to
ensure fair comparison across different security scenarios.
 Expected Outcome: Insights into the relative advantages of AI-driven security solutions
over traditional methods, highlighting opportunities for integrating AI techniques to
enhance existing security infrastructures in fog and edge networks.

· Objective 5: Scalability and Adaptability

 Description: This objective focuses on investigating the scalability and adaptability of

AI-driven security approaches to accommodate diverse network sizes, traffic patterns,
and deployment environments in fog and edge networks. Scalability refers to the ability
to handle increasing amounts of data or devices, while adaptability concerns the ability to
respond to dynamic changes in network conditions and security threats.
 Methodology: Design experiments or simulations that simulate varying network scales
and configurations to evaluate the scalability of AI-driven security solutions. Assess
adaptability by introducing dynamic changes (e.g., network topology changes, new types
of attacks) and measuring how AI models adjust their behavior and performance.

Page 4 of 16
  Expected Outcome: Understanding the limits and capabilities of AI-driven security
approaches in scaling across different fog and edge network architectures, with insights
into strategies for enhancing adaptability to evolving threat landscapes.

Need and Significance of the Research

The research into AI-driven security for fog and edge networks is critical in addressing the
inherent complexities and vulnerabilities of distributed computing environments. As IoT devices
continue to proliferate, demanding real-time data processing and facing resource constraints,
traditional security measures prove inadequate. AI solutions offer adaptive, scalable, and
efficient defenses by leveraging machine learning to detect and respond to threats in real-time,
thereby ensuring robust protection against evolving cyber threats. Furthermore, the integration of
blockchain technology enhances security by establishing decentralized and immutable
transaction records, safeguarding data integrity and authenticity across diverse devices and cloud
infrastructures. This research not only strengthens cybersecurity measures but also supports the
scalability and efficiency required for widespread deployment of fog and edge computing in
sectors such as healthcare, manufacturing, and smart infrastructure, where secure and reliable
data management is paramount for operational success and user trust.

Key Reasons for Research

Complex and Heterogeneous Environments: Fog and edge networks consist of a diverse array
of IoT devices, each with varying computational capabilities and security requirements. The
complexity and heterogeneity of these environments make it challenging to implement uniform
security measures. AI-driven solutions can tailor security protocols to the specific needs of each
device, ensuring comprehensive protection across the network.

Scalability of Security Solutions: Traditional security frameworks often struggle to scale with
the rapid expansion of IoT deployments. AI-driven security mechanisms, on the other hand, can
adapt to increasing volumes of data and devices by leveraging machine learning algorithms that
learn and evolve over time. This scalability is essential for supporting the growth of fog and edge
computing in various industries.

Real-Time Threat Detection and Response: The need for real-time data processing in fog and
edge networks necessitates equally rapid threat detection and response capabilities. AI
techniques such as anomaly detection and predictive analytics can identify and mitigate security
threats in real-time, preventing potential breaches and minimizing damage.

Resource Constraints: IoT devices often operate under significant resource constraints,
including limited processing power, memory, and energy. AI-driven security solutions can
optimize resource usage by processing security data locally at the edge, reducing the need for
constant communication with centralized cloud servers and conserving bandwidth.

Sector-Specific Applications: The research has significant implications for various sectors:

Page 5 of 16
1. Healthcare: Secure and efficient edge computing is vital for applications such as remote
patient monitoring, telemedicine, and real-time diagnostics. AI-driven security ensures the
confidentiality and integrity of sensitive health data.
2. Manufacturing: In industrial IoT applications, secure edge computing can enhance
predictive maintenance, optimize supply chain management, and improve operational
efficiency. AI-driven security mitigates risks of cyber-attacks that could disrupt production
processes.
3. Smart Infrastructure: For smart cities and infrastructure, robust security measures are
crucial for protecting critical systems such as energy grids, transportation networks, and
public safety operations. AI and blockchain technologies ensure reliable and secure data
management.

Supporting Regulatory Compliance: As data privacy regulations become more stringent

worldwide, ensuring compliance with standards such as GDPR, HIPAA, and others is
increasingly important. AI-driven security frameworks can help organizations maintain
compliance by continuously monitoring and safeguarding data flows, identifying potential
violations, and automating compliance reporting.

Advancement of Cybersecurity Practices: The research contributes to the advancement of

cybersecurity practices by introducing innovative AI and blockchain-based solutions. These
advancements not only address current security challenges but also pave the way for future
innovations, establishing new benchmarks for secure IoT and edge computing deployments.

Enhancing User Trust and Adoption: Secure and reliable data management is paramount for
gaining user trust and facilitating the widespread adoption of fog and edge computing
technologies. By ensuring robust security measures, this research supports the broader
acceptance and implementation of these technologies across various industries, driving
innovation and operational success.

Problem Statement

The integration of AI with fog and edge computing presents unique opportunities to enhance
security. However, existing security frameworks are often inadequate to address the complex,
dynamic, and resource-constrained environments of fog and edge networks. This research aims
to develop an AI-driven security framework that ensures data confidentiality, integrity, and
availability while addressing the specific challenges of fog and edge computing.

Research Objectives

The main objectives of this study are as follows:

Develop and Evaluate an AI-driven Security Framework: This objective involves creating a
robust framework that ensures data confidentiality, integrity, and availability in fog and edge

Page 6 of 16
networks. The framework will employ AI techniques such as machine learning and anomaly
detection to dynamically adapt to emerging threats and secure data exchanges across distributed
environments.

Investigate AI Techniques' Impact on Performance and Security: This objective focuses on

assessing how AI algorithms enhance both the performance and security of fog and edge
computing. Metrics such as latency reduction, throughput optimization, and resource efficiency
will be analyzed alongside security improvements to quantify the benefits of AI integration.

Manage User Privacy and Regulatory Compliance: Addressing user privacy concerns and
complying with data protection regulations (like GDPR or CCPA) is crucial. The AI-driven
security framework will incorporate mechanisms for anonymization, access control, and data
encryption to safeguard sensitive information while ensuring legal compliance.

Address Integration Challenges: Integrating AI with existing fog and edge architectures poses
technical challenges such as compatibility issues, resource constraints, and interoperability. This
objective aims to identify these challenges and propose solutions to seamlessly integrate AI
technologies into heterogeneous computing environments.

Mitigate Privacy Concerns: AI-driven security measures can potentially raise privacy concerns,
particularly regarding data collection, processing transparency, and user consent. This objective
involves developing strategies to mitigate these concerns through ethical AI practices,
transparency in algorithmic decision-making, and user-centric design principles.

Compare with Existing Solutions: To assess the effectiveness of the proposed AI-driven
security framework, comparative analysis with existing solutions will be conducted. Evaluation
criteria will include effectiveness in threat detection, efficiency in resource utilization, scalability
across device networks, and adaptability to diverse application scenarios.

Research Questions

 What are the main security challenges in fog and edge networks?
 How can AI be leveraged to enhance the security of fog and edge computing
environments?
 What are the potential risks and benefits of integrating AI-driven security measures in fog
and edge networks?
 How can the proposed AI-driven security framework improve data confidentiality,
integrity, and availability in fog and edge networks?
 What are the privacy implications of AI-driven security measures, and how can they be
mitigated?

Literature Review:

AI-driven security involves the application of artificial intelligence techniques to enhance the
detection, prevention, and response to security threats in networked systems. In fog and edge

Page 7 of 16
computing environments, where data processing occurs closer to the data source (e.g., IoT
devices), AI plays a crucial role in mitigating vulnerabilities and ensuring robust cybersecurity
measures. For instance, AI algorithms can analyze large volumes of data in real-time, detecting
anomalies and potential threats faster than traditional methods relying on manual intervention or
rule-based systems (Gartner, 2021).Fog and edge networks introduce unique security challenges
due to their decentralized nature and reliance on IoT devices with varying levels of security
measures. According to industry reports, the proliferation of edge computing is expected to reach
75% of enterprise-generated data processed outside traditional centralized data centers by 2025
(Gartner, 2021). This shift necessitates enhanced security measures at the network edge to
protect sensitive data and ensure continuous operations in dynamic and distributed computing
environments.

Current Trends and Challenges in AI-Driven Security:

Recent advancements in AI technologies, such as machine learning and deep learning, have
revolutionized network security practices by enabling adaptive and predictive security measures.
Organizations are increasingly investing in AI-driven cybersecurity solutions to address evolving
cyber threats, particularly in IoT and edge computing environments. For example, NVIDIA's
EGX platform enables edge AI processing, allowing organizations to deploy AI models for real-
time threat detection without relying on centralized cloud resources (NVIDIA, 2022).

Literature Gap Identification:

Despite the progress in AI-driven security, there exist significant gaps in literature specifically
tailored for fog and edge networks. Current research often overlooks the unique challenges posed
by edge computing environments, such as limited computational resources, intermittent
connectivity, and diverse IoT device ecosystems. Studies focusing on adapting AI models to
operate efficiently in edge scenarios, while ensuring scalability and real-time responsiveness, are
sparse (Intel, 2021).

Methodologies and Frameworks in AI for Network Security:

Methodologies for applying AI to network security in fog and edge networks include supervised
learning for malware detection, unsupervised learning for anomaly detection in network traffic,
and reinforcement learning for adaptive security policies. These frameworks leverage AI's ability
to learn from vast datasets and adapt to evolving threats in real-time. Practical implementations,
such as Microsoft Azure Sphere, integrate AI-powered security features to protect IoT devices at
the edge, ensuring secure communication and data integrity (Microsoft, 2022).

Case Studies and Empirical Findings:

Empirical findings demonstrate the efficacy of AI-driven security solutions in real-world edge
deployments. For example, Intel Labs' research highlights significant reductions in false positive
rates using AI-driven anomaly detection models deployed at the edge, improving overall security
posture and operational efficiency (Intel, 2021). Case studies of AI-enabled edge platforms

Page 8 of 16
illustrate their capability to detect and mitigate security threats autonomously, minimizing
reliance on manual intervention and enhancing network resilience (Deloitte, 2023).

Integration of AI with Traditional Security Approaches:

Hybrid approaches combining AI techniques with traditional security methods, such as rule-
based systems and signature-based detection, are emerging as effective strategies in fog and edge
networks. These integrated solutions leverage AI's capabilities in pattern recognition and
adaptive learning to complement existing security infrastructures, ensuring comprehensive
protection against sophisticated cyber threats. Evaluating the interoperability and performance
implications of these hybrid models remains a critical area of research and development
(Bonawitz et al., 2017).

Future Directions and Emerging Technologies:

Future research directions in AI-driven security for fog and edge networks include advancements
in model interpretability, privacy-preserving AI techniques (e.g., federated learning, differential
privacy), and edge-native AI frameworks. Organizations are increasingly investing in AI-
powered cybersecurity solutions to enhance detection and response capabilities in distributed
computing environments. These innovations aim to address evolving cybersecurity challenges
while optimizing resource utilization and maintaining data privacy at the network edge (Deloitte,
2023).

Synergies and Challenges:

Synergies: The integration of AI enhances the capabilities of traditional security approaches by

providing real-time threat intelligence, adaptive security policies, and automated response
mechanisms tailored to the dynamic nature of edge networks.

Challenges: However, challenges such as interoperability between AI models and existing

network infrastructures, increased performance overhead due to AI processing requirements at
the edge, and higher maintenance costs for deploying and updating AI-driven security solutions
need to be addressed (Aujla et al., 2019).

Evaluation:

Interoperability: Assessing how AI models integrate with existing network architectures and
security protocols is crucial for ensuring seamless operation and compatibility.

Performance Overhead: Analyzing the computational resources and energy consumption

required by AI algorithms in edge environments helps evaluate their feasibility and efficiency.

Maintenance Costs: Considering the ongoing costs associated with updating AI models,
ensuring cybersecurity hygiene, and training personnel to manage AI-driven security solutions
optimally.

Page 9 of 16
Research Methodology

1. Research Design

To outline the overall approach and design of the research study.This research adopts a mixed-
methods approach combining qualitative and quantitative techniques to comprehensively
evaluate AI-driven security solutions in fog and edge networks. The methodology involves both
experimental simulations and empirical analysis of real-world deployments to ensure robustness
and practical applicability.

2. Data Collection

To specify the sources and methods used to collect data for the studyData collection
encompasses both primary and secondary sources. Primary data will be gathered through
controlled experiments and simulations conducted in a testbed environment mimicking edge
computing scenarios. Secondary data will include literature review findings, case studies, and
industry reports to inform the theoretical framework and validate empirical results.

3. Experimental Setup

To detail the setup and configuration of experimental environments for testing AI-driven security
mechanisms.The experimental setup will involve deploying AI models (e.g., machine learning
algorithms for anomaly detection, deep learning models for image-based security analysis) on
edge computing platforms (e.g., NVIDIA EGX, Azure IoT Edge). Virtualized environments will
replicate diverse network conditions, varying levels of traffic loads, and simulated security
threats to evaluate model performance, scalability, and resilience in real-time edge computing
scenarios.

4. Data Analysis

To explain the methods and techniques employed to analyze collected data and derive
meaningful insights.Data analysis will employ statistical methods to measure key performance
indicators such as detection accuracy, false positive rates, and response times of AI-driven
security solutions. Qualitative analysis techniques, including thematic coding of qualitative data
from interviews or expert feedback, will complement quantitative findings to provide a holistic
assessment of AI's effectiveness in enhancing security resilience at the network edge.

5. Ethical Considerations

To address ethical considerations and ensure compliance with research ethics standards.The
study will adhere to ethical guidelines regarding data privacy, confidentiality, and informed

Page 10 of 16
consent when collecting and analyzing data. Measures will be taken to anonymize sensitive
information and protect participants' identities in case studies or interviews involving
stakeholders or end-users of AI-driven security solutions in fog and edge networks.

6. Validation and Reliability

To ensure the validity and reliability of research findings.Validation will be achieved through
rigorous testing and validation procedures of AI models in controlled environments, comparing
results with established benchmarks and theoretical predictions. Reliability will be enhanced
through repeated experiments, ensuring consistency and reproducibility of findings across
different experimental conditions and scenarios.

7. Limitations

To acknowledge potential limitations and constraints of the research study. Limitations may
include the constraints of simulated environments compared to real-world complexities,
variability in edge computing infrastructures, and potential biases in data collection or analysis.
Transparency in reporting limitations will guide interpretations of research outcomes and
recommendations for future studies.

Proposed Framework

The proposed AI-driven security framework for fog and edge networks will incorporate the
following components:

 AI Algorithms for Threat Detection: Machine learning models will be used to identify and
respond to security threats in real-time, adapting to new and evolving threats.
 Data Encryption and Access Control: Advanced encryption techniques and access control
mechanisms will ensure data confidentiality and integrity.
 Privacy-Preserving Mechanisms: Homomorphic encryption and secure multi-party
computation will be employed to protect user privacy and ensure compliance with data
protection regulations.
 Resource Optimization: AI algorithms will optimize resource usage, ensuring that security
measures do not impact the performance of fog and edge devices.

Diagram :

Page 11 of 16
Implications of the Research

1. Technological Advancements

Discuss the technological advancements and innovations that may result from the
research.Implementing AI-driven security solutions in fog and edge networks can lead to
significant technological advancements. These include:

 Enhanced Security Posture: Improved detection accuracy and response times to cyber threats,
reducing vulnerabilities in edge computing environments.
 Optimized Resource Utilization: Efficient use of computational resources at the edge, minimizing
latency and enhancing overall network performance.

Page 12 of 16
 Scalable Solutions: Scalability of AI models to accommodate growing volumes of data and diverse
IoT device ecosystems, ensuring robust security measures across distributed networks.

2. Operational and Strategic Implications

Explore the operational and strategic implications for organizations and stakeholders. The
research findings will have operational implications such as:

 Cost Efficiency: Potential cost savings through reduced incidents of cyber attacks and operational
downtime, optimizing resource allocation and maintenance costs.
 Strategic Integration: Integration of AI-driven security into existing IT infrastructure and business
operations, aligning cybersecurity strategies with organizational goals and compliance requirements.
 Competitive Advantage: Gain a competitive edge by leveraging advanced AI technologies to
mitigate risks and enhance customer trust in IoT-enabled services and applications.

3. Policy and Regulatory Considerations

Address policy and regulatory considerations related to AI-driven security in fog and edge
networks. The study will inform policy makers and regulatory bodies about:

Data Privacy and Compliance: Ensuring compliance with data protection regulations (e.g., GDPR,
CCPA) and industry standards for secure data handling and processing at the edge.

Ethical Use of AI: Guidelines for ethical deployment of AI-driven security solutions, safeguarding user
privacy and maintaining transparency in AI decision-making processes.

4. Social and Economic Impact

Assess the broader social and economic impact of AI-driven security innovations.Potential
impacts include:

 Job Creation: Opportunities for skilled professionals in AI cybersecurity roles, driving innovation
and expertise in emerging technologies.
 Digital Inclusion: Facilitating broader access to secure IoT services and applications, promoting
digital inclusion and equitable access to technology benefits.
 Resilient Infrastructure: Strengthening resilience against cyber threats in critical infrastructure
sectors (e.g., healthcare, transportation) relying on edge computing for real-time data processing and
decision-making.

5. Future Research Directions

Propose future research directions based on the study’s findings and identified gaps.Areas for
future research may include:

Page 13 of 16
 Advanced AI Models: Development of more adaptive and context-aware AI models for dynamic
edge environments, addressing evolving cybersecurity challenges and edge computing constraints.
 Interdisciplinary Collaboration: Collaborative research initiatives integrating AI with disciplines
such as network architecture, IoT security, and human-computer interaction to advance holistic
approaches to edge security.
 Long-term Impact Assessment: Longitudinal studies to assess the long-term effectiveness and
sustainability of AI-driven security solutions in mitigating emerging threats and adapting to evolving
technological landscapes.

Conclusion

The integration of AI with fog and edge computing holds tremendous potential to enhance
security measures, making networks more resilient to threats and more efficient in data handling.
As the proliferation of IoT devices continues to accelerate, the need for robust security
frameworks that can adapt to the dynamic and distributed nature of fog and edge networks
becomes increasingly critical.

This research aims to develop an AI-driven security framework tailored specifically for fog and
edge networks. The proposed framework will address current challenges such as resource
constraints, latency requirements, and the heterogeneity of IoT devices. By leveraging advanced
AI techniques, including machine learning, deep learning, and anomaly detection algorithms, the
framework will provide proactive threat detection, real-time response capabilities, and adaptive
security policies.

References

Chen, M., Masseglia, F., & Tziritas, N. (2021). AI-Driven Security in Edge Computing: A
Review, Challenges, and Future Directions. IEEE Internet of Things Journal, 8(15), 12433-
12445. doi:10.1109/JIOT.2021.3104984
Li, D., Liu, Y., Li, Y., & Cheng, X. (2020). Blockchain-Based Secure and Privacy-
Preserving Data Sharing Scheme for Fog Computing. IEEE Transactions on Industrial
Informatics, 16(6), 4322-4330. doi:10.1109/TII.2019.2946938
Ren, Y., Zhang, Y., Yan, Y., & Xu, L. D. (2018). On Edge Intelligence: Challenges,
Opportunities, and Solutions. IEEE Intelligent Systems, 33(5), 4-9.
doi:10.1109/MIS.2018.112130859
Liu, Z., Zhou, Y., Zheng, K., Chen, M., & Jia, W. (2021). A Survey on Security and Privacy
of Edge Computing: Potential Solutions, Recent Advances, and Open Issues. IEEE
Communications Surveys & Tutorials, 23(3), 1905-1930.
doi:10.1109/COMST.2021.3057157

Page 14 of 16
Gupta, H., Kaur, M., & Kumar, N. (2020). A Comprehensive Review on Fog Computing:
State-of-the-Art Challenges and Solutions. Journal of Network and Computer Applications,
154, 102574. doi:10.1016/j.jnca.2020.102574
Dinh, T. T. A., Liu, R., Zhang, M., Chen, G., Ooi, B. C., & Wang, J. (2018). Untangling
Blockchain: A Data Processing View of Blockchain Systems. IEEE Transactions on
Knowledge and Data Engineering, 30(7), 1366-1385. doi:10.1109/TKDE.2018.2799738
Yao, H., Wang, Q., Zeadally, S., & Chen, J. (2021). Blockchain-Enabled Secure Computing
in Mobile Edge Networks. IEEE Transactions on Industrial Informatics, 17(4), 2814-2822.
doi:10.1109/TII.2020.2992686
Gelenbe, E., & Lent, R. (2017). Self-aware networks and network security. ACM
Transactions on Autonomous and Adaptive Systems, 12(3), 1-23. doi:10.1145/3078861
Liu, F., Tong, J., Zhang, Q., & Wei, L. (2020). Fog Computing: Focusing on Security and
Privacy Challenges. IEEE Internet of Things Journal, 7(7), 5881-5890.
doi:10.1109/JIOT.2020.2987418
Wang, H., Jiang, B., Chen, J., Han, Z., & Rodriguez, J. (2019). On Deep Learning for Trust-
Aware Edge Computing in Internet of Things. IEEE Internet of Things Journal, 6(2), 1860-
1869. doi:10.1109/JIOT.2018.2867767
Chen, M., Zhang, Y., Li, Q., Mao, S., & Leung, V. C. M. (2018). Deep Learning for Smart
Industry: Efficient and Scalable Framework Using Intelligent Fog Computing. IEEE
Network, 32(4), 70-75. doi:10.1109/MNET.2018.1700231
Qureshi, B. M., Javaid, N., Khan, Z. A., Alrajeh, N., & Guizani, M. (2019). Cognitive Fog
Computing Paradigm: A Survey. IEEE Access, 7, 68164-68181.
doi:10.1109/ACCESS.2019.2917344
Prakash, A., Moran, D., Liu, J., & Lee, Y. (2017). Data Security and Privacy in Fog
Computing: Challenges and Opportunities. IEEE Cloud Computing, 4(5), 16-24.
doi:10.1109/MCC.2017.91
Su, W., Wang, Q., Zhang, Y., Wang, F. Y., & Ren, K. (2020). Blockchain and Artificial
Intelligence for Secure Data Sharing in IoT Systems: A Survey. IEEE Access, 8, 110639-
110652. doi:10.1109/ACCESS.2020.3005764
Liang, X., Zhao, J., Shetty, S., Liu, J., & Li, D. (2019). Towards Secure and Privacy-
Preserving Data Sharing in IoT-Enabled Healthcare Systems. IEEE Communications
Magazine, 57(4), 44-49. doi:10.1109/MCOM.2019.1800735
Wang, S., Zhang, Y., Jin, J., & Wan, J. (2018). Edge Computing: A Potential Enabler of
Internet of Things. IEEE Access, 6, 13856-13869. doi:10.1109/ACCESS.2018.2793819
Wang, Q., Wang, S., Ren, K., Lou, W., & Li, J. (2019). Towards Secure and Privacy-
Preserving Data Sharing in IoT-Enabled Healthcare Systems. IEEE Communications
Magazine, 57(4), 44-49. doi:10.1109/MCOM.2019.1800735

Page 15 of 16
Zhou, X., Wen, S., Li, J., Ren, K., & Lou, W. (2018). Enabling Fine-Grained Multi-Keyword
Search Supporting Classified Sub-Dictionaries over Encrypted Cloud Data. IEEE
Transactions on Parallel and Distributed Systems, 29(8), 1798-1811.
doi:10.1109/TPDS.2018.2792062
Le, L., Hoang, D. T., & Tuan, H. D. (2019). Blockchain for Secure and Efficient Data
Sharing in IoT-aided Smart Grid: A Survey. IEEE Transactions on Industrial Informatics,
15(8), 4620-4629. doi:10.1109/TII.2019.2922603
Ndiaye, M., Taleb, T., & Samdanis, K. (2019). When Blockchain Meets IoT: Towards a
Sustainable Future. IEEE Transactions on Industrial Informatics, 15(6), 3656-3666.
doi:10.1109/TII.2018.2879860

Page 16 of 16