Deep Reinforcement Learning-Empowered Resource Allocation for Mobile Edge Computing in Cellular V2X Networks

Area of Science:

• Vehicular Communications
• Mobile Edge Computing
• Artificial Intelligence

Background:

• Vehicular networks face resource scarcity for extensive V2X computations.
• Cloud computing introduces communication delays and energy issues for vehicles.
• Mobile Edge Computing (MEC) offers a solution by bringing computation closer to vehicles.

Purpose of the Study:

• To develop an optimized computational resource allocation strategy for vehicular user equipment (VUE).
• To minimize the combined cost of energy consumption, computation latency, and communication in V2X scenarios.
• To address the challenges posed by dynamic vehicular environments and limited network resources.

Main Methods:

• Proposed a joint optimization algorithm utilizing the Double Deep Q Network (Double DQN).
• Leveraged deep reinforcement learning for dynamic resource allocation.
• Focused on minimizing a cost function encompassing energy, latency, and communication.

Main Results:

• The Double DQN algorithm demonstrated improved performance over other reinforcement learning methods.
• Achieved approximately 30% better convergence, 15% reduction in defined cost, and 17% increase in speed.
• The algorithm is well-suited for dynamic and low-latency vehicular communication scenarios.

Conclusions:

• The proposed Double DQN-based algorithm effectively optimizes MEC for V2X communications.
• Offers a practical solution for enhancing vehicular computing capabilities while reducing costs.
• Highlights the potential of deep reinforcement learning in addressing real-world vehicular network challenges.