The Study Of Dynamic Policy Based Resource Allocation In Mobile Environments

With the rapid development of wireless technology, the applications ofwireless network are gettingmore and more abroad,the requests of wirelessare also getting more and more. There are not only data services, but alsovideo and audio services in the mobile environments. Because of mobileusers'movement and wireless channel capacity's limitation, providing betterqualityofservicetomobileusersisbecomingmoreandmorecomplexity.The traditional network management does not have a unified standardand pattern, it depends on network manager's experiences or some simplenetwork management protocols to manage the network. The traditionalnetwork management focuses on the network equipments. With the rapiddevelopment and more applications of communication, the complexity andisomerism aremoreobviously,thetraditional networkmanagement is tofacemoreandmoreproblems.In order to satisfy the wireless network's development, using policymechanism to manage the network is necessary. Policy is a set of rules tomanage, configure and control the access to network resources. PolicyBasedNetwork Management (PBNM) is used to simplify network management.PBNM describes and implements the high-level policy to manage the entirenetwork rather than individual equipment. It translates policy intoconfiguration commands and implements these configurations through thenetwork in order that users can manage and configure network equipmentswith the simple commands. PBNM can provide automatic configuration ofnetwork equipments, and ensure end-to-end network management, as well aspredicating and controlling preferential treatment. Network Management canbe simplified while the network is regarded as an entity rather than acollectionofindividualnetworkequipments. IETF (Internet Engineering Task Force) has defined policy rules by thecondition-action form. The rule carries out its actions after triggering theconditions. Network system under the policy control is a piece of statemachine, policies determine equipment's legal state. The general form ofrules is"IF Condition THEN Action". The actions'parameters of staticpolicies are constant which are defined by network managers who accordingto their experiences, requirements of network and so on. Such static policiesareunabletoadapttothedynamicchangesofnetworkstate.In order to solve the disadvantages of managing network with staticpolicies, in this paper we proposed a concept of dynamic policy. Theparameters of dynamic policy's actions are variable, the value of theparameterscanchangeaccordingtothenetworkstates,sowhentheresourcesof network are abundance, the system can offer better service to the mobileusers; while the resources of network are few, the system can offerdegradation service to mobile users. Using the dynamic policies to managethe network resources can provide better QoS to users, allocate networkresources more flexibly and adapt the changes of network state. Usingdynamic policies can improve the performances of network in mobileenvironments.IETF cooperated with DMTF (Distributed Management Task Force)have proposed the PBNM realization frame. They divided the PBNMarchitecture into four major function entities which are PolicyRepository,PMT (Policy Management Tool), PDP (Policy Decision Point) and PEP(Policy Enforcement Point). User can use PMT to editor the policies; ThePMT stores the policies into PolicyRepositoryafter dealingwith the policies;PDP is responsible for the explanation with policy rules and initiate to carryout; PEP carries out the actions of policy rules actually, and feedbacks theresult of carrying out. In addition, IETF has also made communicationprotocol COPS (Common Open Policy Service) which exchanges the policy information, COPS-RSVP and COPS-PR expand respectively thecommunicationprotocol(COPS).In order to adapt the mobile environments, the two entities are added:pre-information cache and policy cache, based on the PBNM architectureproposed by IETF, to improve the real-time character of networkmanagement and utilize the network resources more effectively.Pre-informationcacheisusedtostorethepre-informationwhichhasnotbeenavailable. When the users want to move into theother cells, firstly, theysendout the pre-information to the neighbor cells, when the cell received thepre-information,thePDP estimates the ArrivingTimeofpre-information,andthen decides to deal with the pre-information or not. If the PDP received thepre-information and deal with pre-information immediately, the PDP maychoose unsuitable policies and send them to PEP, these may cause thenetwork resources can not be utilized effectively and mobile users can notconnect the channel normally. The policy cache is used to store the policieswhich are decided by the PDP. When the PDP decided the policies, in ordertoavoidoccupyingthechannel prematurely,the PDP sends thepolices to thepolicy cache. When the mobile users arrived the cell, the policy cache sendsthe policies to the PEP, then the PEP manage the network equipments andnetworkresourcesbythepoliciesdecidedbythePDP.Forthe mobileusers obtainingbetterQoS,the limitedchannel resourcesbeing utilized more effectively, we proposed an approach to allocate channelresourcesbaseddynamicpoliciesinthispaper.First, the calls can be divided to real-time calls and nonreal-time callsaccording to their sensitivity of time delay, and according to the motion, thecallscanbedividedtonewcalls,handoffcallsandongoingcalls.Then, we divide the channel into real-time channel and nonreal-timechannel. Real-time calls can occupy real-time channel, especially when thereis not enough real-time channel capacity, real-time handoff calls can occupy the nonreal-time channel; when nonreal-time channel has free capacity, thesystem provides best-effort servicetononreal-timecalls, but thenonreal-timechannel should release the channel resources when real-time calls want tooccupythe channel resources which are occupied bynonreal-time calls. Thisapproach of dividing channel can keep the real-time calls'priorities, andnonreal-time calls need not wait for too long time, so that nonreal-time callscanachievebetterQoS.After dividing logical channel, the policy system offers differentservices to different types of services according to the channel states. Thepolicy system can also adjust the capacities of real-time channel andnonreal-time channel and the schemes of allocating resources, according tothe call blocking probability, call dropping probabilityand channel resourcesutilizationprobability.An approach of resources management based on dynamic policy isproposed in this paper. In this way, we can make real-time services obtainhigher priority to occupy the channel resources and nonreal-time obtainconnectingthechannel.Usingdynamicpoliciestomanagenetworkresourcescan eliminate errors caused by traditional network management whichfocusesonnetworkequipments,adaptnetworkresourceschangedynamicallyandofferbetterQoStomobileusers.