Search

A novel endpoint admission control scheme (EAC) in IP networks is proposed in this paper. EAC mechanisms are driven by independent measurements taken by the edge nodes on flow of packets injected in the network to probe the source-to-destination path. We dig into the intrinsic relations among accepted load, QoS achieved and the delay variation statistics, and deduced a few theorems which guide throughout the whole admission control study. Our scheme is characterized by two fundamental features. First, it does not rely on any additional procedure in internal network routers other than the capability to apply different service priorities to probing and data packets. Second, the connection admission decision is based on the analysis of the probing flow delay variation statistics. With only two priority levels used, which different from stateful centralized solutions, this scheme can only provide a single level of QoS. To overcome this limitation, it can be used to perform admission control within a DiffServ class to provide different QoS performance for hetero traffic. NS2 simulation results, which focus on IP telephony scenario, show that, despite the lack of core routers cooperation, toll-quality performance figures (99th delay percentiles not greater than few ms per router) can be obtained even in severe overload conditions. Finally, a comparison with an EAC scheme driven by probe losses only, shows that the use of delay variation statistics as endpoint decision criterion is a key factor for EAC effectiveness.

Approximately 40% of all email in the Internet is spam, and its volume is growing rapidly. Blackhole list and mail filter are the main measures to defend the spam at present. But as a matter of fact, the situation of spam flooding tends to be more and more terrible. In this paper, we proposed a new technology of anti-spam based on flows, which could detect spam and abnormal email behaviors in the network, according to the type of email flows and their respective characteristics. First, the impacts of spam are introduced. The conception of flow and some relative definitions are given. Then three types of email flows and their judgment models are presented. And three cases we aimed at to resolve the spam are discussed. Finally, network measurement system, the platform of deploying the anti-spam, and the policy-based scheduling are introduced.

SYN flooding attack is a common method employed in denial of service (DoS) and distributed denial of service (DDoS). It's hard to maintain the effective defense merely by passive defense measures such as monitoring and filtering. To avoid influencing the legitimate service requests and stop attacks at the source, the attack-detection and service-protection must be combined with the trace of the attack source. Having studied and concluded the features of such attacks, we propose a few key metrics to judge if an attack is undergoing. In this paper, the based platform, network measurement system (NMS) is outlined first. Then the detection method is discussed in detail, focusing on the features of service-protection, attack-elimination, and how to trace back the attack source. Finally, we present the experiments of the defending mechanism and analyze their results.