Cognitive Congestion Control for Data Portals with Variable Link Capacity
Network congestion, one of the challenging tasks in communication networks, leads to queuing delays, packet loss or the blocking of new connections. In this paper, a data portal is considered as an application-based network and a cognitive method is proposed to deal with congestion in this kind of network. Unlike previous methods for congestion control, the proposed method is an effective approach for congestion control when the link capacity and information inquiries are unknown or variable. Using sufficient training samples and the current value of the network parameters, available bandwidth is adjusted to distribute the bandwidth among the active flows. The proposed cognitive method was tested under such situations as unexpected variations in link capacity and oscillatory behavior of the bandwidth.