PAW: Priority-based adaptive weighting for software defined video load balancing

This paper presents a novel multipath load balancing algorithm called priority-based adaptive weighting (PAW) for video over software defined networks (SDN). Based on priority differentiation among video packets, PAW adopts a per-packet strategy to distribute ingress video packets over the multiple egress ports of an entry SDN-switch connecting to those output links leading to the same IP destination, where their selection weights are provided by an SDN-controller and adaptive to time varying available multipath bandwidths. The experimental results show that PAW substantially outperforms all the non-priority-based weighting schemes for load balancing, such as adaptive weighting (AW), fixed weighting (FW) and equal weighting (EW).