Vector Offset Based Multipath Scheduling for Differentiated Services

Multipath transmission can effectively integrate path resources and provide high-quality QoS performance for services. However, traditional scheduling algorithms are unable to flexibly identify suitable transmission paths based on service demands due to the absence of support for fine-grained sensing technology. To solve this problem, this paper proposes a vector offset based multipath scheduling algorithm (VOMS) to optimize the adaptability of service demands and path resources. First, the proposed algorithm collects path resources information in real time based on In-band network telemetry (INT) and constructs a multipath resources and service demands joint matrix. Then, VOMS uses the vector offset method to calculate Euclidean distance to rank path scores based on service demands. Finally, scheduling strategies are established according to service demands, prioritizing the selection of single path transmission to meet service requirements. If the bandwidth of a single path is insufficient, VOMS allocates traffic based on sorted paths from highest to lowest until the total filled bandwidth meets the service demand. Experimental results show that compared with the traditional round robin scheduling algorithm (RR) and the bandwidth-based weighted round robin scheduling algorithm (WRR), VOMS can ensure high QoS performance for the service and significantly reduce the packet out-of-order rate.