An AI-enabled lightweight data fusion and load optimization approach for Internet of Things

Area of Science:

• Computer Science
• Network Engineering
• Data Science

Background:

• Internet of Things (IoT) applications generate highly correlated and redundant data due to overlapping sensor node ranges.
• Data processing and transmission in dense IoT networks deplete node energy and overload network gateways and edge servers.
• Heterogeneous edge servers lead to unbalanced loads, causing delays, packet loss, and degraded Quality of Service (QoS) for time-critical applications.

Purpose of the Study:

• To ensure QoS in IoT applications by addressing data redundancy and unbalanced edge server loads.
• To propose a lightweight data fusion approach to eliminate data correlation at the source.
• To introduce a dynamic service migration technique for optimal load balancing across heterogeneous edge servers.

Main Methods:

• A lightweight data fusion approach partitions node buffers to broadcast only non-correlated data.
• A dynamic service migration technique is proposed to reconfigure loads across edge servers.
• Two meta-heuristic algorithms and a migration approach are used to solve the load balancing optimization problem, dynamically adjusting configurations based on server load thresholds predicted by machine learning.

Main Results:

• The proposed data fusion method effectively reduces data correlation at the node level.
• The dynamic service migration technique successfully balances loads across heterogeneous edge servers.
• Experimental results demonstrate the efficiency of the approach for large-scale, densely populated IoT applications, maintaining optimal Gateway-Edge configurations.

Conclusions:

• The integrated approach of lightweight data fusion and dynamic service migration significantly enhances QoS in dense IoT environments.
• The method provides an efficient solution for managing redundant data and optimizing resource utilization in fog computing infrastructures.
• This research offers a scalable and effective strategy for improving the performance and reliability of delay-sensitive IoT applications.