Optimal Joint Scheduling and Congestion Window Estimation for Congestion Avoidance in Mobile Opportunistic Networks

Mobile Opportunistic Networks (MONs) are characterized by frequently changing network topologies and intermittent connectivity due to nodes' limited communication range. These dynamics often lead to constrained buffer resources, causing excessive message drops and inefficient replication strategies, which in turn result in high overhead and performance degradation from buffer congestion and head-of-line blocking. Existing replication-based routing schemes fail to account for these limitations, particularly in considering the impact of contact duration on network performance. In this paper, we develop a congestion avoidance mechanism that addresses these issues by modeling contact duration using a lognormal distribution, derived from both real-world mobility traces and synthetic mobility models. Our approach includes the prediction and detection of buffer congestion, the calculation of optimal congestion window sizes, and the design of an efficient message scheduling algorithm to minimize buffer occupancy and prevent packet drops. This mechanism is applicable to both homogeneous and heterogeneous contact networks. Simulation results on synthetic and real-world mobility traces demonstrate that our proposed scheme significantly reduces overhead while enhancing delivery performance, outperforming existing approaches.