Researchers have developed a new distributional reinforcement learning algorithm called DSAC-T to optimize energy and resource allocation in heterogeneous mobile edge computing (MEC) systems assisted by active beyond-diagonal reconfigurable intelligent surfaces (BD-RIS). This approach models return distributions, enhancing policy stability and feasibility, particularly under reward heterogeneity. In simulations, DSAC-T achieved an 81.67% feasibility ratio and a rapid online decision time of 0.0267 seconds per scenario, outperforming other baseline algorithms. AI
IMPACT This research could lead to more efficient resource management in edge computing environments, improving performance and reducing energy consumption for AI-driven applications.
RANK_REASON Academic paper detailing a new algorithm and its application. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
