Deep Reinforcement Learning-Based Power Allocation for Minimizing Age of Information and Energy Consumption in...

Area of Science:

• Wireless communication systems
• Internet-of-Things (IoT)
• Signal processing

Background:

• Multi-input multi-output and non-orthogonal multiple access (MIMO-NOMA) systems enhance capacity and efficiency for real-time IoT applications.
• Age of information (AoI) is critical for timeliness in real-time IoT data.
• Base station (BS) controls sampling and power allocation, impacting AoI and energy use.

Purpose of the Study:

• To optimize sample collection commands and power allocation in MIMO-NOMA IoT systems.
• To minimize both Age of Information (AoI) and energy consumption.
• To leverage deep reinforcement learning (DRL) for optimal power allocation strategies.

Main Methods:

• Proposed an optimal power allocation strategy using deep reinforcement learning (DRL).
• Simulated the proposed DRL-based power allocation against other algorithms.
• Utilized successive interference cancellation (SIC) at the BS for signal decoding.

Main Results:

• The DRL-based optimal power allocation achieved lower AoI and energy consumption.
• Demonstrated significant performance improvements compared to genetic algorithm (GA) and random algorithms.
• Achieved a reward reduction of 6.44% compared to GA and 11.78% compared to random algorithms.

Conclusions:

• Optimal power allocation is crucial for minimizing AoI and energy consumption in MIMO-NOMA IoT systems.
• DRL provides an effective method for achieving optimal power allocation.
• The proposed DRL approach offers superior performance over existing methods for real-time IoT applications.