Font Size: a A A

Research On Policy-Based QoS Management In Mobile IP Network

Posted on:2009-05-25Degree:DoctorType:Dissertation
Country:ChinaCandidate:D WeiFull Text:PDF
GTID:1118360245963214Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the development of wireless technology and mobile communications, itbecomes necessary for NGN to support mobility. Policy-based NetworkManagement (PBNM) is a new kind of network management scheme. Itemphasizes management automatization and uses the execution of policies torealize management. So it can better meet the demands of variability andtimeliness of mobile IP networks and will promote its QoS management. Torealize the policy-based QoS guarantee techniques, IETF defined threeinformation models including PCIM, PCIMe and QPIM. These models togetherachieved the description of QoS guarantee for static networks. It is required tobuild corresponding QoS guarantee policy information models for applyingPBNM to mobile IP networks. But the models are missing now. So firstly, thisdissertation puts forward a policy information model, called MQPIM, to describetheQoS guarantee techniques under mobileenvironments inorder tosupport QoSin mobile IP networks. Then it creates a new policy-based registrationmanagement mechanism for mobile IPnetworks and extends the COPS-MIPv6 tosupport mobile terminals'policy-based registration management in mobile IPv6networks. Thus it becomes feasible to support policy-based network managementin mobile IPnetworks. Because of the features such as mobilityand timeliness ofmobile IP networks, it is demanded for policy-based management systems tocreatenewpolicies dynamicallyorbuilddynamic parameters accordingto outsideenvironments. So,this dissertationputs forwardthedefinitionofdynamic policiesin policy-based network management systems, uses product space algebra ofconditional events and one point-coverage random set to update dynamic policiesand then builds two models for dynamic policies updating. Finally, an improvedalgorithm based policy actions of gradating is put forward to find potential conflicts and then provides necessary supports for dynamic prolicies'conflictsdetection and resolution. The main contents and conclusions in this dissertationcanbesummarizedasfollows:1.After analyzing the features of kinds of topical mobile networks, it furtheranalyzed all techniques used for QoS guarantee under mobile environments andascertained the special and fundamental characters of mobile environments whichare different from static environments. Then combined with PCIM, PCIMe andQPIM information models defined by IETF, it advanced a mobile QoS policyinformation model(MQPIM) supporting the definition, description and store ofQoS policy under mobile environments。This information model extended theQPIM(RFC3644) using object-oriented method and provided a kind of generaldescription language used for QoS policypresentation under mobile environments.At the same time, it gave out the detailed definition of MQPIM, including theaccurate definitions of classes, the causes of these definitions and its detaileddirections of usage. Based on the traditional QoS, MQPIM took the QoS undermobile environments into account, supported mobile QoS guarantee, supportedcontrolling the mobile procedures by policies, controlled the usage of wirelessresources and described QoS guarantee techniques under mobile environments.In the background that QoS guarantee techniques are not good enough andstandards are not ready now, MQPIM made it come true to describe the QoSpolicies in mobile IPnetworks. Also because MQPIM put forward some importantfactorswhichshouldbeconsideredfortheissues aboutQoSguaranteeinmobileIPnetworks, it pointed out some necessary reference directions for studies on QoSguarantee techniques in mobile IP networks. In addition, with development ofMQPIM needed to be enriched and updated continuously to meet the demands ofQoSpoliciesdescriptionfornewQoSguaranteetechniques.2.Aimingatthepolicy-basedregistrationmanagementinmobileIPnetworks,it firstly analyzed the procedure of policy-based registration management inmobile IPv4 networks and the COPS-MIP, the communications protocol betweenPEP and PDP. Because there is no corresponding standard for communicationsprotocol between PEP and PDP about policy-based registration management inmobile IPv6 networks, a policy-based registration management mechanism and anew protocol COPS-MIPv6 are put forward to support policy-based registrationmanagement for mobile terminals in mobile IPv6 networks. The new protocol defined some objects such as context and mobile options and fulfilled the correctencapsulation for messages of mobile IPv6. In order to complete policy-basedregistrationmanagement, it also definedcommunications interfaces ofCOPS-MIP,namely COPS-MIP communications modules for PDP and PEP respectively, allobjects and message-making methods of COPS-MIP, an interface to receivemobile IP message and a mobile IP user interface used to deal with mobile IPmessages. COPS-MIPv6 is a new protocol extended form COPS-MIP and it isuseful to solve the communication problems between PEP and PDP inpolicy-based registration management. It provided fundamental guarantee for thestudyonpolicy-basednetworkmanagementinMobileIPv6networks.3.AimingatthedemandsofmobilityandtimelinessofmobileIPnetworks,itset up policy-based events space model based on policy core information modeland its extensions defined byIETF, provided the definition of dynamic policies inthe policy-based network management system and studied the characters ofproduct space algebra of conditional events and one point-coverage random set.Combined with the features of policy-based network management, two updatingmodels for dynamic policies are set up, one is PS-CEA-based dynamic policiesupdating model and the other is one point-coverage random set-based dynamicpolicies updating model. In the first model, it put forward a similarity-metricalgorithm for rules and then set up a new policies updating model which canacquire global optimization by taking the rule as a whole. In the universalframework for dealing with uncertain and fuzzy information, the second modeltook the rule similarity as the events with constant probability in the PS-CEA,made information converging of several groups of related rules and then acquiredthe balance between the factors such as the arbitrariness of linear weight eventsand the difficulties of setting up non-linear weight events model. By comparingwith rule-based reasoning system, it found that the model can deal with thequalitativelyandquantificationallyinformationdescriptioninpoliciesprimlywiththe same time complexity. Also the model can fully use the information such asexperience and knowledge, make the updated dynamic policies meet the QoSdemandsofmobiletrafficandmakethemostofnetworkresources.4.Because of the emerging of dynamic policies, there are some dynamicfeatures in the policies conflicts detection. So it studied the conflicts evaluationmethods based on policy actions and put forward an improved algorithm based policyactions ofgradatinginordertofindthepotential conflicts ofpolicyactionsand provide necessary supports for conflicts detection and resolution of dynamicpolicies. Combined with characters of dynamic policies, it analyzed policy-basedQoS management environment, ascertained the action attribution set and decisionattribution set of policy, and mined the relationship between combination ofpolicy actions in the set of similar policies and network performance indices byusing rough set-based attribution reduction method. In detail, it analyzed thedifferentiation matrix-based attribution reduction algorithm in rough set theory,improved the algorithm by combining with the setup of posets relationship ofaction attributions in the policy database, then reduced the decision informationtable formed by policy action attributions and network performance indices byusingthe algorithm and form series of decision rules. These decision rules helpedto avoid the creation of dynamic policies which maybe conflict with otherspotentially and reduce the load of maintainment of network performance. Thesedecision rules also helped to select the one most compatible with existing policysetinmorethanonechoiceofpolicieswhendynamicpolicytriggeredandtriedtooptimizethenetworkperformance.
Keywords/Search Tags:MobileIP, Policy-based Network Management, QoS, dynamic policy updating, product space algebra of conditional events, onepoint-coverage random set, roughset, attribution reduction
PDF Full Text Request
Related items