wild card : metropolitan area networks

49
Wild Card : Metropolitan Area Networks By Greg Standerfer

Upload: nemo

Post on 23-Feb-2016

28 views

Category:

Documents


0 download

DESCRIPTION

Wild Card : Metropolitan Area Networks. By Greg Standerfer. IEEE Standard for Local and Metropolitan Area Networks: Overview and Architecture. Sponsored by the LAN/MAN Standards Committee. LAN vs. MAN. LAN – Local Area Network. MAN – metropolitan area network. Single building, lab, etc - PowerPoint PPT Presentation

TRANSCRIPT

Wild Card : Metropolitan Area Networks

Wild Card :Metropolitan Area NetworksBy Greg StanderferIEEE Standard for Local andMetropolitan Area Networks:Overview and ArchitectureSponsored by theLAN/MAN Standards Committee

LAN vs. MANLAN Local Area NetworkMAN metropolitan area networkSingle building, lab, etcRun and operated by single organizationCouple of blocks to whole cityRun by multiple organizationsBoth are peer to peer communication networksBoth have moderate-to-high data ratespacket-based communication capabilities

Standards for IEEEFollow the Open Systems Interconnection (OSI) Basic Reference Model (OSI/RM) And cover the lowest two layers : data layer and application layerData layer is divided into two parts : LLC (logical link control) and MAC (medium access control)

LLC LayerConnected to MAC layer through a single MAC service access point (MSAP) Communicates in three different ways :unacknowledged connectionless-mode (type 1),connection-mode (type 2)acknowledged connectionless-mode (type 3)

MAC LayerCommunicates through the use of packets called MAC framesPrimary functions are:frame delimiting and recognition Addressing of destination stations,transparent data transfer of LLC error protection direct access to the physical layerinterconnection layer devicesNeed to have all interconnection devices operate correctly and transparently for large amounts of end to end systemsphysical layers have repeaters or hubs MAC layer has bridges Connect multiple access domains Network layer has routers

Unique IdentifiersLAN/MAN networks allow for unique MAC addresses,group addresses, and protocol identifiersThis is done by assigning Organizationally Unique Identifiers (OUIs) to each end systemOUIs = 24 bits and assigned by IEEELast 24 assigned by organization

IEEE 802.16* WirelessMAN* Specification Accelerates WirelessBroadband AccessD.J. Johnston

Broadband accessHas been huge delays in the US to get broadband access.In 2002, the US only held 17 percent of the total broadband subscribersReasons:Are out of reach of DSL servicesAre not part of a residential cable infrastructureThink its too expensive

WirelessMANDesigned to fix all these problems and get more broadband to the US customersBenefits:Do not need towers, (can be put on tall buildingsEasy to installEasy for customer to link up with Base stationScalable Cheaper than wires.Phone companies are slower802.16 standard 2400 ISPs in the United States serving 6000 markets all with their own standards and expensive setupMakes it very expensive to setup large wireless networks in this type of marketNeed a common ground of technologies.802.16 sought to combine the same technologies throughout all wireless ISPs carriersCan handle 60 businesses in one area with T1-type connectivity

802.16 (continued)This new technological advance allows for a whole new set of opportunities to arrisestandard 802.16a has been developed with a frequency of 2-11 Ghznon-line-of-sight operation Mesh networkMesh network allows subcarriers to forward messages through other subcarriers to a base stationAllows users mobilityWiMaxWorldwide Interoperability for Microwave Accessfixed and fully mobile internet access and refers to implementation of 802.16Created in June 2001Was created closely with the IEEE and European Telecommunications Standards Institute (ETSI)The ETSI version is HIPERMANHopesThere is a lot of hope for this new WirelessMAN technology to decrease the costs of having broadband internetSeveral companies are starting to make larger pushes with this technology. (CLEAR)Only time may tellFinding an Effective Sustainable Model for a WirelessMetropolitan-Area Network: Analyzing the Case of PittsburghJ. M. Peha, B. E. Gilden, R. J. Savage, S. Sheng, B. L. Yankiver

WMANBecoming cheap enough and financially logical enough for cities to implement.Cities have a lot of issues to balance in order to decide whether or not to have a WMAN network and how implement it.ConsiderationsTo maximize the areaMaximize competition for better pricesMinimize subsidiesTo ensure it is financially worth it, and can make a profit

ModelsMonopoly One for the whole cityFacilities based competition separate entities running their own networks that compete against each other, and overlapWholesale-retail one network for the city, with wholesale in charge, and allows retailers to sign up users, manage accounts and payers while user the services the wholesale providesOpen-competition done by default where all ISPs people have a free for all for who gets interconnectivity

City PoliciesPolicies to affect first year - One time cash payment for the first year for the initial construction and infrastructure needsPolicies to affect annually reducing annual costs or revenue for the vendorsCity can be the biggest default subscriber. Need to have to the city utilize the network as much as possible to ensure it will not failRevenue and Cost projectionsNeed to compare the city to other cities that are in similar situations Can make estimates based on thatmean cost of a WMAX is $111,000 per square milePittsburgh = 55.5 milesAround $6.5 million for first yearProjected Revenues

Projected Costs

Model Comparisons

What Pittsburgh should do?Have a WMANvertically integrated monopoly Or a citywide wholesaler with competing retailersDYNAMIC RESOURCE ALLOCATION IN OFDMAWIRELESS METROPOLITAN AREA NETWORKSSYED HUSSAIN ALI, KI-DONG LEE, AND VICTOR C. M. LEUNGTHE UNIVERSITY OF BRITISH COLUMBIAOFDMAOrthogonal Frequency-Division Multiple Accessa physical layer specification for wireless MANssupport 2-11 GHznon-line-of-sight operationmesh operationMesh OperationAllow subscriber stations (SS) to use other subscriber stations as relay station to relay station to a base stationDone when a base station channel is in poor conditionLowers the cost of communicationComplicates the network resourcesmultihop routing (MHR) Resource Allocation problemsCapacity planning (CP)Call admission control (CAC)dynamic subcarrier assignments (DSA)adaptive power allocation(APA)

Capacity planningProcess of determining the productioncapacity needed by an networkstatic optimization problemEach base station is allocated a bandwidth based on its previous usageDone with system offline and very time consumingProblems:effect of group mobility users on QoS the effect of fluctuation in channel gains on QoSNote: that blocking a new call is more favorable that dropping one

Call admission control When each call from a subscriber comes in, it allocates bandwidth based on how much CP has allocated for the base stationDone in real timeTrouble occurs at edges of stationsHard to determine when to pass off SSs to new BS

Dynamic subcarrier assignments/ adaptive power allocationOFDMA assigns subcarrier time slots for each carrier to a base station = DSAallocates the carriers of the frameAPA = how much power goes into the frame at the time when the connection is madeBoth need to occur at the same time and in real timeMost important resource allocation problemsNetwork Types1. PMP mode= point to multipointbase station has a central rolecontrols CAC and DSA/APA since it is in charge of the bandwidth allocation2. Mesh modeMore difficult since it is not centralizedDSA may be implemented to change or reuse frequencies APA can be implemented to not interfere with other subscribers.

Optimization solutionsassumed continuous data rates and an infinite number of subcarriersProblems:SSs closest to the BSs dominate the timeDifferent APAs have performance improvements are marginal compared to SNRPractical SolutionsAssume finite number of subcarriers and discrete data ratesAPA is the same for all SSsHungarian method -O(n^3) iterative heuristic method - O(N)

Opportunistic Fair Scheduling for the Downlink of IEEE 802.16WirelessMetropolitan Area NetworksMehri Mehrjoo, Mehrdad Dianati, Xuemin (Sherman) Shen, Kshirasagar Naik

PMPDeals with problems with the second mode of networks in WMANIn PMP, BS coordinates with more than SSs downlink and uplink directionsIEEE does not specify how a BS should schedule its transmissions Need effective scheduler Good SchedulerBS that forwards real time and non real time messages Prevents the bottleneck that occurs at the BS for downlink (downlink has much higher demands)Needs to non real time downloads, which are delay tolerantAllows for real time to work more effectively which are not delay tolerant

How network workscross layer design - sends packets based on its maximum capacityMessage postponed if bad channel of communication Causes SSs closer to BSs to dominateNeed a fairness model to prevent SSs from starvingSS associates with BS, and BS assigns it a timeframeUL and DL are done with time division multiplexingAnd share the same bandwidth

Scheduler solutionFairness is key SNR can be random and fadeBut too much fairness will cause all time to be spent on the ones that do not work.Need utility function Adds quality of service into the allocation equationutility function for non real time service is concave function while real time is delay-based function

TestingBS placed in center of equally spaced SSsThree models performed demonstrating low amount of traffic to heavy trafficThree modelsRound robin Just basic sharingOpportunistic - Straight picking the best connectionsOpportunistic fair

ConclusionYes Opportunistic had the best throughput which was expected, but Opportunistic fair had a better throughput than round robin on both sides. Opportunistic is also much more fair than the other two algorithms, especially at a