TTE Network Scheduling Directional Search Low-Latency Traffic

The patent changes the sorting parameters from rate-based to latency-based ordering. Instead of sorting VLs by transmit rate (highest first), the algorithm now sorts by latency constraints (tightest first). This parameter change allows low-rate, low-latency VLs to be prioritized correctly, resolving the contradiction between maintaining high scheduling throughput and properly handling traffic that violates the inverse rate-latency assumption.

The patent implements bidirectional search by allowing the algorithm to search for available bins in both forward and reverse directions through the bin sequence. When a fixed-direction search fails to find valid time slots for VL reservations, the algorithm reverses the search direction. This inversion approach maintains search simplicity while significantly improving reliability by ensuring valid schedules are found even when bins are occupied in patterns that block unidirectional search.

The patent applies machine learning and demand prediction techniques in advance to analyze traffic patterns and predict future bin utilization requirements. By performing this analysis beforehand, the system builds a knowledge base that guides the scheduling algorithm's bin selection process. This preliminary action improves scheduling efficiency for complex traffic patterns while managing algorithm complexity by separating the predictive analysis phase from the real-time scheduling execution phase.