Research On Congestion Control In Content-centric Networking

Along with the rapid development of Internet users, the demand of network bandwidth and content personalization are becoming increasingly high. Today Internet is significantly shifting towards a network model where contents play a central role; Content-Centric Networking(CCN) is a promising one among them. With respect to TCP/IP, the transport model of CCN is connectionless with a unique endpoint at the receiver. CCN uses the content name, instead of IP address, to identify the packet. CCN attempts to name content with userfriendly, structured, effectively location-independent names. It has the natural ability of flow control to make the network traffic balance through appropriate forwarding strategy. Another feature of CCN is the possibility to embed storage capabilities into the network. The new features of CCN may cause some new transport problems and congestion may occur when aggregate conversations arrive in excess, so we need some extend control schemes.This thesis starts from CCN’s network architecture and briefly describes the factors which cause network congestion in this new architecture. We propose some flow and congestion control schemes in CCN, then using the open source simulator based on NS-3——ndnSIM to prove its validity. Based on a window-based flow control and a rate-based shaping mechanism, we propose several optimizations to tackle the drawbacks of these two scenarios. First, we propose an approach of combining random early detection with window-based flow control and prove that this combination can get better network throughput. For the rate-based shaping mechanism on router, we introduce a feedback signal for congestion to obtain more accurately congestion information and more quickly respond to congestion. On the other hand, this mechanism prevents the decline of link utilization caused by miscalculation. We propose a weighted round robin mechanism for each shaping queue and prove it strengthens the fairness between each flow. In addition, for the case of large number of shaping queues, we restrict the number of shaping queues according to its popularity and prove this mechanism reduces the network delay caused by shaping.In this thesis, we also focus on DoS/DDoS in next generation Internet architecture. CCN is exposed to Denial of Service(DoS) attack by malicious users sending a large number of requests rapidly. And this kind of attack may cause network congestion. We prove DoS attacks in CCN may affect the quality of service of legitimate users. We analyze two typical types of DoS attacks in CCN, respectively, and introduce the generation and some countermeasures of CS poisoning attacks and Interest flooding attack in detail. Then we propose a congestion-driven cache management policy from the perspective of cache. We prove the effectiveness of this mechanism to deal with CS polluting attacks through simulation. On the other hand, we use a satisfaction-based weighted round robin mechanism according to statistics on router. We prove that this mechanism can effectively inhibit the attack traffic and can get better stability. These mechanisms for CCN to apply in the next generation network architecture have a guiding significance.