The Research On Congestion Control Strategies In Named Data Networking

Named Data Networking is one of the most promising future Internet architecture to replace today's TCP/IP networks because of its connectionless,ubiquitous cache,and hop-by-hop forwarding characteristics.Compared with the TCP / IP,NDN focuses on named data rather than its location(IP address),and the Data becomes a first-level entity of NDN.The emergence of new network architecture will bring about new research contents,and congestion control is one of the focuses.Although NDN is consumer-driven,an Interest brings back a Data transmission mode,which can maintain the self-balance of network traffic to a certain extent;on the other hand,the unpredictability of network traffic,the limitation of network resources and the multi-source and multi-path transmission mode of NDN result in a large amount of redundant data in the network,making congestion still an inherent property of NDN networks.Because of the new network architecture and transmission model of NDN,the congestion control mechanisms of traditional TCP/IP network perform poorly in NDN.In addition,the window-based congestion control algorithm is widely used in NDN,but it cannot reach the optimal throughput because it cannot determine the optimal sending window size;Furthermore,NDN routing is separated from the forwarding plane,and inherently supports acyclic multi-path forwarding,which brings new solutions to NDN's congestion control.However,most of the existing multi-path forwarding schemes have problems such as lagging response to congestion,difficulty in expansion,and high complexity.All these situations force us to design some new congestion control mechanisms to solve these problems.Starting from the NDN network architecture,this thesis firstly introduces the related Congestion technologies of this new network architecture,and then designs a Precise Feedback-based Explicit Congestion Control(PFECC)scheme based on the existing research results of Congestion Control.PFECC designs the congestion control strategy from both the consumer and the intermediate node.Consumer adjust the sending window of each flow accurately and in time according to the explicit feedback carried in the data packet,so as to quickly converge to the target rate and improve network throughput.The PFECC downstream router monitors the queuing delay trend of the upstream router's queued data packets to split the interest traffic,timely and effectively use the link with lower load for multipath forwarding.When the link state returns to normal,the multipath forwarding strategy can automatically switch traffic to the best path for forwarding,which can effectively avoid congestion and improve the throughput of the network.Finally,the proposed scheme was simulated on the ndnSIM platform and compared with the other relevant algorithms.The results show that the PFECC strategy can effectively reduce the packet transmission delay,treat each flow fairly and converge to the optimal throughput faster.