Wireless Medium Access Control Protocols - PowerPoint

Document Sample
Wireless Medium Access Control Protocols - PowerPoint Powered By Docstoc
					Wireless Medium Access
   Control Protocols

A Survery by Ajay Chandra V.
Gummalla and John O. Limb
   Survey
   Distributed vs. Centralized Networks
   Wireless MAC Issues
       Low Power Sensor Nodes
   Random Access
   Guaranteed Access
   Hybrid Access
Introduction Cont’d.
   Distributed MAC Protocols
       Distributed Foundation Wirelesss MAC
       Eliminate Yield – Non-Preemptive Priority
        Multiple Access (EY-NPMA)
Introduction Cont’d.
   Centralized MAC Protocols
       Random Access
           Idle Sense Multiple Acces (ISMA)
           Randomly Addressed Polling (RAP)
           Resource Auction Multiple Access (RAMA)
       Guaranteed Access
           Zhang’s and Acampora’s Proposals
           Disposable Token MAC Protocol (DTMP)
Introduction Cont’d.
    Hybrid Access
         Random Reservation Protocols (RRA)
         Packet Reservation Multiple Access (PRMA)
         Random Reservation Access – Independent
          Stations Algorithm (RRA-ISA)
         Distributed Queuing Request Updated Multiple
          Access (DQRUMA)
         Moble Access Scheme based on Contention and
          Reservation for ATM (MASCARA)
Introduction Cont’d.
      Dynamic Slot Assignment ++ (DSA++)
Distributed Wireless Network
   ad hoc network
   No central administration
   Multi-hop wireless networks
   Wireless Sensor Nets
Centralized Wireless Network
   Last Hop Network
   Very common
       Corporate, Academic, and Cellular uses.
   Has a controlling Base Station, with
    variable intelligence
       Wireless Access Point
       Cellular Tower
Wireless MAC Issues
   Half-Duplex
       No Collision Detection
       Uplink and Downlink must be multiplexed
   Time Varying Channel
       Reflection, Diffraction, and Scattering
       Different version of signal are superimposed on
        each other
       Multipath Propagation
       Coherence Time = time signal strength changes
        by 3dB
Wireless MAC Issues Cont’d.
   Burst Channel Errors
       Higher BER
       Errors occur in long bursts
       Link Layer retransmission based on
        immediate ACKs
Wireless MAC Issues Cont’d.
   Location Dependent Carrier Sensing
       Hidden Nodes: Node A doesn’t know Node B is
        also talking to BS
       Exposed Nodes: Node A knows node B is talking,
        but doesn’t know that it will not affect Node A’s
        conversation with BS
       Capture: Node A and B are both transmitting to
        BS, but since Node A’s signal strength is stronger
        Node A’s transmission is used no collision is
Random Access
   Random Access is based on a “Talk
    whenever you want” way of thinking
   Collisions are resolved by a contention
    resolution algorithm
   Distributed Networks
Guaranteed Access
   Access to medium is scheduled
   Round Robin
   Master/Slave (Polling)
   Tokens
Hybrid Access
   Melds best qualities of Random and
    Guaranteed Access
   Request-Grant mechanisms
   Requests are Random Access, and once
    reserved transmission is guaranteed
   Random Reservation Access
   Demand Assignment
       Distributed Foundation
      Wireless MAC (DFWMAC)
   802.11 Standard
   4-way exchange: RTS-
   No ACK causes sender
    to retransmit
   No CTS causes
    exponential backoff
   RTS and CTS contain a
    NAV which details how
    much data is to be sent
Elimination Yield – Non-Preemptive
Priority Multiple Access (EY-NPMA)
   Sense channel for time to send (TTS) 1700
    bits, if clear, then send
   If busy, N slots; When done listen again
   If still busy, abort; Else listen again, and if
    not busy then transmit until finished
Idle Sense Multiple Access
   Carrier Sensing and Collision detection are
    performed by the BS
   When medium is idle BS broadcasts idle
    signal (IS)
   Nodes with data send
   If collision BS cannot decode signal, does not
    send ACK and broadcasts IS again
   Otherwise BS sends ACK/ISA (ISA)
   Efficiency is improved by using small
    Reservation packets
        Randomly Address Polling
   Nodes with data broadcast orthogonal
    “codes” simultaneously
   BS receives all codes, using a CDMA receiver
   BS then polls each code
       All nodes with that code transmit
       If only one node the BS sends ACK
       More than one node with code causes BS to send
   Reservation RAP supports nodes with
    streaming traffic
     Resource Auction Multiple
         Access (RAMA)
   Each node has and N-bit ID and transmits it,
    in contention phase
   BS then echos back ID it heard bit-by-bit
   Once a node receives a bit it did not transmit,
    it drops out
   Since BS does an OR operation on received
    IDs then node with highest ID always wins
Zhang’s Proposal
   BS polls each node for data, round
   Node responds with data request, or a
    keep alive if queue is empty
   BS then polls each node that responded
    with a data request
       Disposable Token MAC
          Protocol (DTMP)
   Improves on Zhang’s proposal
   When polling nodes BS indicates if it
    has data to send to nodes
   If no data, then remain silent
   Otherwise send short message
   Transmit any data to send
   Channel is assumed to be reciprocal
Acampora’s Proposal
   Poll, request, data phases
   BS polls each node, if the node has
    data to sends it responds
   The BS the broadcasts this nodes ID so
    that all nodes know the order in which
    to send
   BS then polls nodes each node in turn
    for its data
Various Proposals
Random Reservation Protocols
   Uplink is time slotted
   Each slot large enough to carry one voice
   Downlink is broadcast channel
   Nodes use random access to request
    reservations for data to send
   BS enforces a policy of reservations
       Stream Reservation
       Complete BS scheduling
    Packet Reservation Multiple
          Access (PRMA)
   A node with a back-logged voice packet transmits with
    probability p
   If successful, reserves that slot for following packets
   Data is similar, though no reservations are made
   Different access probabilities are used for voice and data
   Introduction of data packets into voice only network
    decreases efficiency
   Improvements include limited data reservations, separating
    voice and data channels (FRMA), separating request and
    data channels (PRMA++)
   Centralized PRMA uses scheduling to achieve QOS
      Random Reservation Access –
     Independent Stations Algorithm

   BS polls a subset of all nodes
   Subset is defined by the probability of a
    single transmission in a slot is
   BS uses channel history to compute
Distributed-Queuing Request Update
      Multiple Access (DQRUMA)
   Uplink and Downlink are duplexed
   Uplink has request channel and packet
       Request channel is for contention requests
       Packet channel is for data (and piggyback new
        contention requests)
   Downlink has 3 messages: ACK for current
    slot, transmission permission for node to use
    next uplink slot, and data to the nodes
   Better than RAMA and PRMA
       Mobile Access Scheme based on
     Contention and Reservation for ATM
   Frame consists of three periods: broadcast,
    reserved, and contention
   Broadcast informs nodes of structure of
    current frame and scheduled uplink
   Reserved period consists of downlink data,
    and uplink data as defined in broadcast
   Contention is random access and used to
    send new requests to BS
    Dynamic Slot Assignment++
   MAC on uplink is TDMA
   Both uplink and downlink are slotted
   Each downlink slot contains some data and a
    MAC message
   MAC message contains ACK for transmission
    on previous uplink slot and a reservation for
    next uplink slot
   BS collects all requests and schedules uplink
Comparison Summary
   QoS guarantees are not suited to Random
    Access protocols because delay cannot be
   Demand Assignment protocols are best suited
    to multimedia applications
   Random Access lends itself to large networks
   Polling protocols are efficient only for smaller
   TDD protocols perform poorly at high data
    rates due to increase in switching