Neural scheduler for real-time packet switching

To resolve transmission conflict in space-division packet switches, it is necessary to find a conflict-free switching schedule with a fixed and short duration. A neural scheduler is proposed for real-time conflict resolution in space-division packet switches. This neural scheduler has smaller mean and variance of the number of iterations compared to existing ones, and can guarantee that P [required number of iterations ⩽n] ⩾P* where n and P* are the requirements specified by the user