Enhanced recurrent attention-deep Q learning with optimal node constrains and effective penalty based model for data...

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Effective data transmission scheduling is crucial for wireless sensor network (WSN) performance and resource optimization.
• Existing scheduling methods often struggle with dynamic conditions, energy efficiency, and network interference.

Purpose of the Study:

• To propose a novel scheduling method, Recurrent Attention-Deep Q Learning with Optimal Node Constraints and Effective Penalty (RA-DQL-ONC&EP), for WSNs.
• To enhance data transmission scheduling by dynamically considering energy consumption and network interference.

Main Methods:

• The RA-DQL-ONC&EP algorithm integrates a penalty-based model, optimal node constraints, and recurrent attention techniques.
• Dynamic scheduling of data transmission tasks is performed, optimizing for energy efficiency and reliability.

Main Results:

• The proposed RA-DQL-ONC&EP achieved a 91.21% success rate for dependable data transport.
• Achieved a low delay rate of 1.99% and demonstrated significant energy savings of 70% compared to other models.
• Throughput analysis showed a 72% throughput over 1,000 time steps, indicating high efficiency.

Conclusions:

• RA-DQL-ONC&EP is a promising approach for improving data transmission scheduling in WSNs, offering enhanced reliability, efficiency, and energy conservation.
• Optimized scheduling supports critical applications like environmental monitoring, healthcare, and smart cities, contributing to societal well-being.
• The algorithm's efficiency promotes cost savings and resource conservation, making it a socially responsible choice for WSN management.