Blockchain-Enabled Equilibrium for Low-Carbon Computing Power Routing: A Tri-Level Stackelberg Approach

This study develops a tri-level Stackelberg framework for coordinating task routing, carbon-aware pricing, and GPU allocation across geographically distributed edge-computing networks. Regulators determine carbon taxes, a platform operator prices services across heterogeneous data centers, and users select computing destinations under uncertainty. A blockchain-based transparency mechanism reduces information asymmetry in perceived congestion and carbon costs, while a computing-capacity-wall model represents nonlinear throughput degradation under server overload. Routing equilibrium is computed with a logit stochastic-assignment method and continuous averaging. Experiments on a 24-node network with eight data centers demonstrate better spatial load distribution, lower latency, and greater social welfare, while revealing a GPU congestion trap at intermediate task ratios.

2026 41st Youth Academic Annual Conference of Chinese Association of Automation (YAC), 2168-2173, doi: 10.1109/YAC71005.2026.11615518