Research Of Policy Actions' Conflict Evaluating Method In Policy Based Network Management

Along with the unceasing development of network communication'sscale and its cover scope, network management technology also unceasinglyweeds through the old to bring forth the new. The policy based networkmanagement has liberated the manager from the numerous and diversedisposition and the maintenance work, simultaneously also has solved theextra worries for the expanded application scale of the network. Thedevelopment and the wide spread application of the mobile communicationtechnology has proposed a new topic to the network management, that is, weneed to guarantee the quality of service of the mobile users and enable theysuccessfully rapidly switch between the network region. If part of resourcesare always reserved for the moving users, it must create the resources waste.As the result, to make the application of network resources rational needs theintroductionofthedynamicresourcemanagementmechanism.The topic"Policy Based Dynamic Resource Management and Assignmentin Moving IP Switching"is proposed under this application background. Thetopic is divided into four main areas: To get forecast information of arrivingmoving IP; according to the forecast information, to generate and trigger thedynamic policies; conflict detection and digestion of dynamic policies; todesignthestoragemethodofdynamicpolicies.The author of this paper mainly participated in the second part --generating and triggering module of the dynamic policy mechanism. In thispart, in order to satisfy the frequent retrieval and matching need in policygenerating and trigger process, we use the condition event algebra theorybased on the product space. In this method, the actions and conditions ofpolicy is separately saved in the independent product space, and these twoparts are related through a mapping function, which enables the policy still tomaintain a complete whole. Moreover, a compatible distinction method basedon the simple point stochastic collection is used through the policy trigger. Inthe entire research process, regarding the formulation of forecast informationpolicy, one part of it needs to completely regenerated, another part mainlyseeks policy which can match in the existing policy stores. Certainly thisprocess must avoid conflicting strategy formation as far as possibly, this alsomayberegardasatriggerprocess. Discovered in the topic discussion process, regardless of in the staticstrategy system or the dynamic strategy system or the static and dynamicunionstrategysystem,thepoliciesinthepolicystorehouseareahugedatabasecontent.Andtheretrieval,matching,updateandconflictdetectionandsoonofthese policies will all face a very high time and space complex question.Thereupon the realization of the intellectualization of policy managementsystem will be imperative. So in view of this main direction o, this moduleproposes a research method based on the policy actions conflict appraisalsystem. The main idea is to build the relation between the policy actions andthe network performance index, excavate the connection rules between theactions get from the similar kind of policy gathers through the policystorehouseandthenetworkperformanceindex,formaseriesofpolicy-makingrules, and build corresponding memory structure which is advantageous forthe search and renew. As a result, we can use these decision-making rules inthe process of the generating or trigger policy as a kind of assistance andjudgment tool. On the one hand, such policy-making rules can play a role inguiding the formation of dynamic policy, avoid to product these existed latentconflict actions in generating period, which can reduce the burden of thenetwork performance maintenance. On the other hand, when policytriggering,faced with such a wide choice of a number of decision-making rules, we canselect the policywhich has best compatibilitywith the existing policies gatheraccording to such policy-making rules, as far as possible optimized networkperformance. Third, create a storage structure through unceasinglyconsummates such policy-making rule, names for the time being as "the latentconflict storehouse", can provide a reference tool for the disposing andavoidingforcollisiondetection.The excavation of the connection rules mainly uses the attribute reductionmethodbasedontheroughsets.Themainworkofthispaperisasfollow:(1) Analysis the QoS management environment based on the policymanagement, thought with the dynamic policy's characteristic of moving IP,confirmtheanalysisobjectsthattheactionattributegathersandpolicy-makingattribute gathers of policies, and withdraws the data through the data structureoftheproductspaceconditioneventalgebra.(2)Through the analysis of the attribute reduction method based on the S-matrix in the theory of rough sets, improve the algorithm integrate theestablishment regarding the action attribute sub-order relations in the union policy database, and using those algorithm carries on reduction of the actionattribute gather of policy and the policy-making information table whichformed by the network system performance index and finally obtains thepolicy-makingrule.(3) Showing the feasibility of above method by an example, and carry ontheanalysisoftheresulttoshowsitsusability.Because at present the experimental environment has not completed, thecontent of this paper is mainly based on theory analysis showing, the dataselected in this paper unavoidably has one-sidedness and subjective factor, itneedsthefurtherdiscussionresearch.(1) The separate question of attribute's successive value in policy-makingtable. The data this paper used are less, therefore it only used the subjectivedivision method. When carries on the analysis to the entire policy database,regarding the separate of each attribute value also needs the massivemanagerial experience and the data analysis. The solution of this questiondirectlyrelatesthefinalreductionresult.(2)The data analysis of this paper all is withdrawed under the samenetwork condition of supposition premise. In other words, each item of recordvalue in the policy-making table takes the value on the supposition networksituation paused in the new policy production time, so all are static data. Butthe actual situation is that the network condition unceasingly changes andrenews. Therefore, to enable our experimental results have the persuasivepower and the confidence level, we can improve this flaw in the after workthrough unceasingly accumulates process. In other words, we put the resultsunder some kind of different environment, if the frequency of this result inother environment appears higher, it shows this result credibly goes fast in abigger way. By this way we can formulate a threshold value of credibility.Afterachievedsuchthresholdmeantthisresultcouldbetrustedandused.Because the ability is limited, the question and the one-sidedness in thepaper are unavoidable, welcome the readers to give the criticism and thesuggestion.