Docstoc

Distributed Control Algorithms for Service Differentiation in Wireless

Document Sample
Distributed Control Algorithms for Service Differentiation in Wireless Powered By Docstoc
					          Distributed Control Algorithms
           for Service Differentiation in
            Wireless Packet Networks


                                    Michael Barry
                                 Andrew T. Campbell
                                    Andras Veres



EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    The paper…
        Investigates service differentiation in wireless packet networks
        Proposes solution based on 802.11 DCF
              VMAC (Virtual MAC)
                     Estimates MAC – level statistics of the channel( delay, delay variation,
                      packet collision, packet loss)
              VS (Virtual Source)
                     Utilizes MAC to estimate application-level service quality
        Demonstrates through simulations that globally stable state may
         be achieved if the proposed algorithms are applied to admission
         control
        Discusses distributed service level management scheme built on
         top of the introduced algorithms

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Motivation
        TCP/IP will be the glue for all applications in mobile
         environments
        Desirable that wireless architecture supports the
         same quality assurances as the wireline Internet
        Wired internet – 2 main approaches.
        1st approach
              Take circuit switching and extend with datagram service
              Very strict control over wireless/wireline resources
              Very rigid, complex


EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Motivation (cont’d)
        2nd approach ( which this paper is based on) :
              Only minimal control and signalling
              Should accommodate many different (unforseen) wireless
               applications
              Good example : 802.11
                     Does not guarantee anything but the best-effort service (In
                      either DCF of PCF mode)




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Motivation (cont’d)
        Base on DCF because PCF is good, but …
              Not currently supported by most wireless cards
              Cooperation with DCF (which is used now) can lead to poor
               performance
              Distributed control (DCF) is more efficient for real-time
               services then the centralized one (PCF).




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Brief DiffServ Introduction
        Concept of Service Differentiation
              Differentiating among services provided by a
               network to different types of customers and
               applications
              IP-layer service differentiated service examples:
                     value-added internet services
                     VPNs
                     application hosting/outsourcing
                     packet telephony


EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Brief DiffServ Introduction (cont’d)
        Proposed and spearheaded by IETF DiffServ task
         force
        Designed to help networks satisfy QoS req’ts.
        Allow different classes of service to be provided to
         streams on a common network infrastructure.
        One of two major models used in QoS
              The other one is Int-Serv
        Reservation-less (as opposed to Int-Serv)
        No state in routers is needed (as opposed to Int-
         Serv)
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Brief DiffServ Introduction (cont’d)
        Push major state and work to the very edges of the
         networks.
              Fan-in and forwarding speeds of flows are smaller
        Packets carry their state in a few bits (DS CodePoint)
        Flows are policed and marked at the edge (when
         they enter the network)
        To support different classes of IP service :
              packet classifiers
              forwarding/per-hop-behavior
              traffic conditioning policies

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Brief DiffServ Introduction (cont’d)
        Services are built by applying rules
              Rules for how packets are marked initially
              Rules for how marked packets are treated at boundaries.
        Three elements work together to deliver a Diffserv
         service:
              Per-Hop Behaviors (PHBs)
                     Deliver special treatment to packets at forwarding time
              Traffic Conditioners
                     Alter packet aggregates to enforces rules for services
              Bandwidth Brokers (Policy Managers)
                     Apply and communicate policy
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Brief DiffServ Introduction (links)
        This is where I found info used in these slides
              http://www.qosforum.com/docs/faq/
              http://www.ietf.org/html.charters/diffserv-charter.html
              ftp://ftp.netlab.ohio-state.edu/pub/jain/courses/cis788-97/integrated_services/index.htm
              ftp://ftp.netlab.ohio-state.edu/pub/jain/courses/cis788-99/h_6qos/sld001.htm




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Approach to providing differentiation
        MAC layer must be differentiation-aware to be able to
         provide any kind of differentiated services in mobile
         environments
              Propose differentiation-aware MAC based on DCF
        Differentiation at radio level not sufficient to provide
         proper differentiation for different traffic types
              Propose VMAC (virtual mac) and VS (virtual source) which
               monitor MAC – level load.
        Base admission control on the estimates.


EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    DCF
        Distributed Coordination Function
        Covered in the EE206A notes
        CSMA/CA
        RTS/CTS for hidden-terminal
        Sense channel for DIFS then tx, else backoff and
         sense again
        ACKs are used



EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    DCF – backoff strategy
        Central to this paper
        Tbackoff = Rand(0,CW) * Tslot
        Tslot is slot time (contention window is broken into
         slots)
        Tbackoff initializes the timer. Timer is decreased only
         when nobody is tx’ing. Otherwise it is frozen.
        Timer is decreased every time the medium idle for
         DIFS (distributed interframe space) interval
        If no ACK after tx: double CW until you reach Cwmax

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    What should diffserv-enabled MAC be?
        Distribute available radio resources between classes
         ensuring that different classes receive different level
         of service
        Adaptive and robust
        Decentralized
        There has been work done to achieve this, but
         proposed solutions have drawbacks, which are
         corrected (obviously) by this paper.



EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Backoff timers as means of differentiation
        Idea: change the CW values to reflect priority.
        Even if collisions, CW values changed @ same rate
              CWmin – lower for higher priority traffic (smaller wait)
              Cwmax – also lower for higher priority traffic. Needed to
               handle time intervals with high congestion
   Drawback:
         • No explicit guarantee of level of differentiation
         • Only relative timings, and even those are based on
         random number generators.

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Simulation and results
        Ns-2 used with wireless extension by Monarch group
        Mix of best effort TCP and better-then-best-effort
         voice connections
        Vary CWmin for both traffic classes
        Increasing CWmin significantly delays TCP packets,
         and only slightly affects voice traffic
        Choice of CWmin significantly affects voice, but not
         best-effort traffice
        The above two are confusing, so here is the graph…

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Simulation - graph




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Simulation - robustness
        Previous slides prove that differentiation is achievable
        Can it be maintained over range of traffic conditions?
        Simulate increasing levels of traffic up to channel
         saturation.
              Yes. Delay goes up for both priority levels, but separation is
               maintained.
              TCP is not starved even at saturation point. That’s due to
               statistical and non-deterministic nature of service separation.




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Simulation – robustness (cont’d)




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – WHY?
        You cant exactly plan the wireless channel
        If you cant plan – adapt through measurement
        Most applications want absolute bounds, not relative




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – HOW?
        Free capacity of the channel may be estimated by
         measuring idle channel time after DIFS
              Remember, nodes are required to sense channel for DIFS
               before even attempting transmissions in DCF
        Algorithms passive, operate in parallel to “real” MAC
         algorithms
        VS and VMAC are related: VS algorithm is actually a
         combination of Virtual Application, Interface Queue,
         and a Virtual MAC (VMAC)


EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – HOW? (cont’d)
        Virtual Application generates packets
        Virtual packets timestamped and placed in the queue
        Scheduling on wireless channel: much as one would
         for real channel
              Sensing
              Backoff
              Collision detection (if you would tx a virtual packet on a busy
               medium)
        One difference from the real MAC: packets are not
         actually transmitted
        All aspects of real MAC are emulated (retries, etc)
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – HOW? (cont’d)
        “real” node would detect collision when it’s RTS timer
         would go off without receiving a CTS
        Virtual nodes detect collisions if they sense a channel
         busy, when they are ready to transmit a virtual
         packet
        Virtual nodes enter backoff procedure after the time
         equal to RTS timer of a real node
        MAC delay on no collision:
              D = tdefer+ tRTS+ tCTS+ tpacket+ tACK+ 3tSIFS+ 3 tau
              tau – maximum propagation delay

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS - Performance
         There is an argument which suggests that Virtual
          MAC is absolutely not resource intensive.
         (virtual and simulated MAC delay)




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – Performance (cont’d)
        Average Delay of a new voice source




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    VMAC and VS – Performance (cont’d)
        Delay variation for a new voice source




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Virtual Delay Curves
        Another important part of the VS algorithm
        Average delay of virtual packets if the VS algorithm
         generates packets at the rate of Prate
        Example of a tradeoff: at the same bitrate
              Reduce application delay by increasing packet rate
              Yet, higher packet rates load channel more
        Applications can make use of this to shape their
         traffic
        Represented by a function d(Prate), where Prate is
         packet inter-arrival time
        Data bitrate is constant Psize * Prate = const
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Virtual Delay Curves (cont’d)
        Mobile host or base station runs VS algorithms with
         several Pratevalues in parallel
        From delay values obtained from VS you may
         construct a delay (and/or delay variance) curve
        From the curve a base station or a mobile host can
         choose optimal packet rate and packet size so that
         the application experiences minimum delay and delay
         variance



EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Admission Control
        Based on VS and VMAC
        Natural location – at the base stations
        For better performance, mobiles should have that as
         well (since radio channel properties may be different
         in mobiles then in base-stations)
        Based on the idea that it’s better to drop then to
         delay past bound packets belonging to better-then-
         best-effort traffic
        Again, studied through simulation

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Admission Control (cont’d)
        Aggregate rates of traffic in simulation (notice
         burstiness of TCP)




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Admission Control (cont’d)
        Estimated delays by VS algorithms in base stations
        Few base stations
         stop admitting traffic
         when load goes way
         high. Some are in
         continuous accept.
         This is due to overlap-
         ping areas of coverage
        Bursty traffic handled
         well
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Service-level
    Management in a Wireless Network
        What does this have in common with the rest of the
         paper?
              Topic of admission control and the word “distributed”




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Service-level
    Management in a Wireless Network
        Mobile moves into the service area, and performs
         handoff you need to perform admission control.
        Base station needs to know SLA
              Service Level Agreement
              “Contract” between provider/user specifying (for example)
               what bitrates the user will supply, and what kind of
               performance the network can guarantee
        So, you need to transfer the SLA context somehow
        It also seems as if this is the very first time DiffServ
         actually kicks in. Before, nothing said proposed
         schemes could be made to work with Int-Serv

EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Distributed Service-level
    Management in a Wireless Network                                                              (cont’d)

        First time the mobile enters the network make it
         obtain an SLA from some centralized control agent.
        SLA is transferred as a “token”
        SLA token is encrypted, and base-stations have keys
        Once Mobile Node obtains SLA it will submit it to
         base stations upon entering the cells.
        This way base-stations need not communicate, and
         are able to perform admission control locally



EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    Critique?
        Did not see too many things I did not like
        Admission control in base-stations – does it contradict DiffServ
         principles of pushing decisions to the edges?
        Seems performance would be impacted by virtual mac although maybe
         nothing major. If placed in base-stations, there will be much more then
         20 (as the graph shows) nodes
        They argue its always better to drop packets then delay them for
         better-then-best-effort traffic. Should this not be left up to
         applications? Again, maybe contradicts the earlier “robustness”
         comment?
        Same statistical / non-deterministic nature that “does not allow TCP
         traffic to be starved” may very well starve it in some cases (or near-
         starve it)
        Distributed Service Level Management… does not seem to be very well
         integrated into the paper
EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks
    References
        The EE206A slides (Spring ’01)
        http://www.en.polyu.edu.hk/~cmsp/Events/doc/tutorial_QoS.pdf




EE206A – SPRING 2001 – Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:9/30/2012
language:English
pages:36