Low-Latency Oriented Joint Data Compression and Resource Allocation in NOMA-MEC Networks: A Deep Reinforcement...

Area of Science:

• Computer Science
• Electrical Engineering
• Network Engineering

Background:

• Mobile edge computing (MEC) faces communication pressure and resource constraints.
• Terminal devices have limited battery capacity and computational power.
• Dynamic channel conditions and task requirements pose challenges.

Purpose of the Study:

• To propose a resource allocation optimization method for MEC systems.
• To minimize total task processing latency under practical constraints.
• To integrate data compression and non-orthogonal multiple access (NOMA) technologies.

Main Methods:

• Joint optimization of computational resource allocation, task offloading, and data compression ratios.
• Formulation of an optimization model to minimize total task processing latency.
• Development of a softmax deep double deterministic policy gradient (D3PG) algorithm for dynamic environments.

Main Results:

• The proposed softmax D3PG algorithm effectively optimizes resource allocation, task offloading, and compression ratios.
• The method successfully minimizes total task processing latency while adhering to constraints.
• Simulation results show superior performance compared to benchmark algorithms.

Conclusions:

• The integrated approach significantly reduces communication pressure and resource constraints in MEC.
• The proposed algorithm enhances convergence speed and task processing efficiency.
• This method offers a viable solution for optimizing MEC systems with limited resources.