Research Of Dynamic Policy Updating Mechanism In Mobile IP

With development of the technology of wireless and mobile communication, it's necessary to provide mobile in the next generation network. The mobile IP network will be the principal part of the future information network.It's important for mobile IP network to guarantee the QoS in wireless network. When the mobile host handoff occurs, how to guarantee the quality of service is a key issue. Policy-Based Network Management (PBNM) is an effective management scheme.As the wide application of mobile IP technology,how to allocate resource reasonably and ensureQoS(Quality of Service) of workflow becomes of great importance. Now, many scientific institutions and researchers take part in the research work of mobile prediction, and put forward some solutions, such as Shadow Cluster , LZ-based , PPM , etc. While import the PBNM(Policy Based Network Management) technology to the mobile environment, therefore produce dynamic policy through prediction of mobility and required resource of moving host, and trigger these policy before the moving host's arrival, and this completes the dynamic management of mobile network, also bring about the result of reasonable resource allocation and insurance of mobile business's QoS. Since the predicted information is uncertain and fuzzy, what's more, the processing methods for uncertain information and fuzzy information are different, therefore, how to form an unified processing framework that can handle both of these two kinds of information means a lot to the fusion of their information. At the same time, how to generate dynamic policy for resource allocation according to predicted information is the key to the problem of dynamic resource allocation in the process of IP switching.Aim at the problem of the QoS of occurrence of mobile IP handoff, policy management system need dynamically create policy or form dynamic policy parameters as the environment changes. So how to create policy and form dynamic param0eters is important in the process of mobile IP hands off.This paper builds policy event space model based on the project assignment issue, policy kernel information model and it's extendibility. According to analyzing the policy spring mechanism of mobile IP hands off, we build a policy spring model and policy spring flow chart.Restricted by technology,knowledge and experience, the decision system can only get incomplete, uncertain and fuzzy information instead of complete information for decision making; at the same time, reference system in being can not avoid information lost when making decisions, which make the accuracy of decision fall greatly.This paper investigates the features of procduct space algebra of conditional events and one-point random set theory, combined with the features of policy-based network management , and we set up two dynamic policy update model for resource allocation, which can not only combine quantitative numerical information and qualitative experiential conditional information closely, but also can make the updated information to satisfy the QoS of mobile business as well as take full advantage of present network resource.the one model is PS-CEA-based mobile IP dynamic policy updating model, proposes a rule comparability measurement arithmetic in this model, this arithmetic is different from many existing rule comparability measurement arithmetics, it takes rules as a integer and builds a integer optimization policy updating model ; the other model is one-point random set theory-based mobile IP dynamic policy updating model, it deals with indeterminate information and fuzzy information uniformly, proposes using rule similar degree as constant probability event of procduct space algebra of conditional events, syncretizing many interrelated information. This idea is a tradeoff among linear weight event which is arbitrary , and nonlinear event whose design is very difficult ,and so on. According to compareing with rule-based consequence system: With the same time complexity, this model can take knowledge study effectively, and can make the best of the known knowledge and experience to make more exactitude decisions; and makes the updated policy can meet the mobile traffic QoS and utilize the network resource adequately. What's more,this kind of method can be used in many other intelligent decision making areas, like: distributed network intrusion detection.One common problem in specifying rules for policy-based networks is the conflict detection problem, where rules may contradict or duplicate each other.Because mobile IP system comprise large collections of heterogeneous and mobile devices, services and applications context, and these mobile devices carry oneself rules which is used to execute in conceate environment can from one locateon to another freely,so that policy conflict probability of mobile IP system is more than other system. We study ECA and it's extendibile mothod to conflict detection and condlict resolution proplem base on dynamic policy model and policy.Procduct space algebra of conditional events and one-point random set theory, and the outcome of the paper can use in high intelligent decision making areas,for example: Data Mining,Condlict Detection,Iterative Estimative, Distributed Network Intrusion Detection;What's more, We can combine Byesian Network,Nenual Network,D-S Evidence Theory into an unified process framework,which make advantage use of these theory dealing with uncertain and fuzzy information and Itegrted Decision, make decision making more precise.The investigations are needed to be continued about that the performances evaluation of policy updating model and applications of policy updating model, and how to use procduct space algebra of conditional events and one-point random set theory to build more complex models.