부산대학교 도서관

The quantum annealing machine manufactured by D-Wave Systems is expected to find the optimal solution for QUBO (Quadratic Unconstrained Binary Optimization) accurately and quickly. This would be useful in future applications where real-time calculation is needed. One such application is traffic signal optimization. Some studies use quantum annealing for this. However, they are formulated in unrealistic settings, such as only crossroads on the map. Therefore, we suggest a QUBO, which can deal with T-junctions and multi-forked roads. To validate the efficiency of our approach, SUMO (Simulation of Urban MObility) is used. This enables us to experiment with geographic information data very close to the real world. We compared results with those using the Gurobi Optimizer in the experiment to confirm that quantum annealing can find a ground state. The results show that the quantum annealing cannot find the ground state, but our model can reduce the time that vehicles wait at a red light. It is also inferior to the Gurobi Optimizer in calculation time. This seems to be due to the D-Wave machine's hardware limitations and noise effects, such as ambient temperature. If these problems are solved, and the number of qubits is increased, the use of quantum annealing is likely to be superior in terms of the speed of calculating an optimal solution.