Router Queue Management for Mixed Traffic

Document Sample
Router Queue Management for Mixed Traffic Powered By Docstoc
					Congestion Control in the
Internet with Mixed Traffic
Sources and Heterogeneous
Access Networks
    Candidacy: Yi Pan (ypan@ics.uci.edu)
    Committee Members:
    Tatsuya Suda (Chair)
    Wei K. Tsai
    Magda El Zarki


                                           1
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
 Introduction
 Solutions: overview
 Wavelet de-noising scheme in the core
  network
 Smooth handoff scheme at the edge
 Conclusion



                                          2
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   The Future Internet
     Multi-service       network
          Carries traffic from more and more applications
             Data transmission: FTP, HTTP
             Streaming applications: VoIP, RTP/RTCP streams

     Ubiquitous       access network
          Integrate many different wireless access networks
           to the Internet
               Different wireless access networks: WiFi, GPRS, WiMAX



                                                                    3
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Challenges for congestion control
     In   the core:
          Application traffic can be feedback controlled (responsive
           traffic) or not (non-responsive traffic)
                Bursty non-responsive traffic causes high queue fluctuation and
                 bursts of packet losses
                Most responsive traffic is long-term TCP traffic
     At   the edge:
          Different wireless access networks have different available
           bandwidth
                Transition of an active application session between different
                 wireless networks often causes transmission rate disruption




                                                                                 4
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
                                                       Web servers
                                        File servers            Media servers
                                                                                       Non-responsive
                                                                                       Traffic Sources
                                    AQM router


                                Core Network



                                                             Packet loss

                                                                                   Responsive
                              Different bandwidth                                  Traffic
  Wireless Access Network 1




                                                                                                5
                                                       Wireless Access Network 2
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
 Introduction
 Solutions: overview
 Wavelet de-noising scheme in the core
  network
 Smooth handoff scheme at the edge
 Conclusion



                                          6
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Solutions
     Wavelet  De-noising Scheme for AQM routers
      in the core network
          Classify different traffic patterns and reduce the
           queue fluctuation caused by non-responsive traffic
     Smooth     handoff scheme at the end-systems
          Smooth transition of data paths as well as the
           transmission rates between heterogeneous
           wireless access networks


                                                                7
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Traffic classification for AQM routers in the core
    network
     Existing     methods
          Classify and identify different traffic patterns through packet
           header examinations
                Multiple layers of protocol headers in the packets are examined
                 to identify different application flows
          Apply different policies on different traffic flows identified
                Packets identified as bursty non-responsive traffic are often
                 dropped



                                                                                 8
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Limitations:
        Deep packet inspection may be prohibited by
              Network management policies and security protocols (IPSec in VPNs)
        Flow identification is not scalable to large number of traffic flows
        Per-flow application traffic patterns may not be indicated by protocol
         headers
              E.g. Rate controlled UDP flows are also responsive traffic, while HTTP
               short transactions on TCP are bursty
        Increasing cost in packet header examination with increasing link
         speeds
              Need complex hardware in the data path of a router to perform line-
               speed packet header examination
        Dropping packets in bursty traffic may impair the performance of
         important applications such as HTTP web browsing


                                                                                        9
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   New features of the proposed wavelet de-
    noising scheme
     Classification
                   of responsive and non-responsive traffic
      based on periodic samples of incoming traffic volume
          By analyzing transmission rate reduction of incoming traffic in
           response to AQM packet loss
                No packet header examination
          Non-responsive traffic is filtered and bypass the AQM queue
                Filtering : a virtual AQM queue length is maintained only
                 corresponding to responsive traffic volume
                Bypassing: available buffer in excess to the virtual AQM queue
                 length is used to allow non-responsive traffic to pass through
                 the router



                                                                              10
 Congestion Control in the Internet
 with Mixed Traffic Sources and
 Heterogeneous Access Networks
 Contributions
    No header examination for each packet
            Avoid restrictions on packet header examination by network
             management and security protocols
      Scalable to number of traffic flows in the network
            No flow identification required
      Not relying on protocol headers to identify application traffic
       patterns
      Scalable to increasing link speed
            No data plane hardware for complex packet processing
             required
            Scheme implemented at control plane of the router
      No additional packet losses when extra buffer is available



                                                                          11
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Solutions
     Wavelet  De-noising Scheme for AQM routers
      in the core network
          Classify different traffic patterns and reduce the
           queue fluctuation caused by non-responsive traffic
     Smooth     handoff scheme at the end-systems
          Smooth transition of data paths as well as the
           transmission rates between heterogeneous
           wireless access networks


                                                            12
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Handoff schemes in the wireless access networks
       Existing handoff schemes
            Only transfer data paths from one access point to another
            Do not consider bandwidth differences in different wireless
             access networks
       Issue
            Handoff between two different wireless access networks can
             cause disruption in transmission rate
                   Transmission rate on the old data path may be too high/low on the
                    new data path
                   Switching to a new data path may require stop the current
                    transmission rate and re-establish a new transmission rate
                   Results in packet loss or low throughput in applications
            Example application suffering from the rate disruption: Real-
             time streaming applications

                                                                                        13
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   New features of the proposed smooth
    handoff scheme
     Use multiple paths to reach a single mobile
      node
          Assign different mobile IP addresses (COAs) to different
           paths reaching a single mobile node
     Exploit different amounts of bandwidth on
      multiple paths to a single mobile node
          To reduce or prevent a packet loss due to hand off
          To increase throughput for the mobile node



                                                                      14
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
 Contributions
      Allow smooth transition of both data paths and the
       transmission rates during handoff
            Using multiple paths during handoff avoids delay in switching
             data paths
            Establish transmission rates on multiple paths during the
             handoff avoids transmission rate disruption during handoff
      Allow optimal usage of available bandwidths on multiple
       paths
            Different available bandwidth on multiple paths are used to
             optimize the video quality during handoff



                                                                             15
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
 Introduction
 Solutions: overview
 Wavelet de-noising scheme in the core
  network
 Smooth handoff scheme at the edge
 Conclusion



                                          16
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   The proposed de-noising scheme



                         De-noising
                          Scheme          Non-responsive traffic
                                          (bypass the AQM queue)



                 Responsive traffic

                                 Traffic entering the AQM virtual queue

                                                                   17
         Congestion Control in the Internet
         with Mixed Traffic Sources and
         Heterogeneous Access Networks
      Wavelet De-noising scheme overview
         Traffic
                estimator
         De-noising filter


                                                                Estimated noise

                                                                   De-noised
                                                                   traffic to AQM




Estimate periodic cycle and average packet loss corresponding
                                                                              18
to responsive traffic
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Wavelet De-noising scheme overview
     Traffic estimator
        Estimate the periodic cycle of responsive traffic
                Most responsive traffic follows TCP AIMD mechanism, resulting
                 in periodic rate changes
           Estimate the AQM packet loss rate corresponding to
            responsive traffic
     De-noising filter
        Removing non-responsive traffic bursts
                Non-responsive traffic bursts occurs in time scales different
                 from the dominating periodic cycle of responsive traffic



                                                                                 19
      Traffic estimator
      Estimate       the periodic cycle of responsive
        traffic                                              Periodic cycle of responsive traffic
             Time shift δtk in
             traffic traces      200




                                 150




                                 100




                                 50



Strong correlation:
                                  0


Incoming traffic comes                 0              1000           2000          3000




                                                                Shifted Incoming Traffic u(t+δtk)
                                                                                                 4000          5000          6000




back to the peak rate in δtk                0.2


                                           0.18



after a peak in AQM packet                 0.16


                                           0.14



loss rate                                  0.12


                                            0.1


                                           0.08


                                           0.06


                                           0.04




                                                                                                                                           20
                                           0.02


                                             0
                                                  0           1000          2000          3000          4000          5000          6000


                                                                       AQM packet loss rate p(t)
Traffic estimator
    Algorithm
         cross-correlations between the incoming traffic and AQM
          packet loss rate are calculated
            corr k   xcorr  pt , u t  t k  , t k  2 k , k  0... log 2 n

          Packet loss rate        Incoming traffic               Time shift length

         The cycle lengths are decided by the strongest cross-
          correlation
                        Tcycle  2i , if maxcorrk   corr[i]




                                                                                           21
          TCP traffic estimator
          Estimate         AQM packet loss rate for responsive
           traffic            200




Linear increasing speed:      150
                                                                               Average TCP rate
                                                                               reduction =
                              100




                               50




                                0
                                    0   1000   2000   3000   4000     5000   6000


                                               Incoming TCP Traffic
                              0.2


                             0.18


                             0.16




                                                                    Nloss: Number of packet loss in
                             0.14


                             0.12




 Average Packet Loss pavg     0.1


                             0.08
                                                                    Tqbusy
                             0.06


                             0.04


                             0.02


                               0
                                    0   1000   2000   3000   4000    5000    6000


   Queue busy cycle Tqbusy
                                               AQM packet loss rate
                                                                                                  22
         Traffic estimator
               The average queue busy cycle length is then


               That gives us the conclusion (Theorem 2.4.1)
                     The AQM packet loss corresponding to responsive traffic is


Queue busy ratio ≈1                          Observed packet loss in queue busy period



                          Link capacity       Observed queue busy period




                                                                                   23
      Traffic estimator
           Simulation results with HTTP-type noise
              2                                                                                 2
           N loss                                                      r  Nloss            
         pavg  C 2                                         ploss                        
                                                                      pavg  Cln k  Tqbusy 
                                                                                           
                      1600



                      1400



                      1200



                      1000
                                                                      The slope is the estimated ploss
                      800


                      600



                      400



                      200


                                                                      2
                        0
                             2   4   6   8   10   12   14    16     Tqbusy

Our estimation on ploss is stable under HTTP-type noise (i.e. close to the line
corresponding to the TCP configuration in the simulation)
                                                                                                    25
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Wavelet De-noising scheme overview
     Traffic   estimator
          Estimate the periodic cycle of responsive traffic
          Estimate the AQM packet loss rate corresponding to
           responsive traffic
     Wavelet      De-noising filter
          Removing non-responsive traffic bursts
               Non-responsive traffic bursts occurs in time scales different
                from the dominating periodic cycle of responsive traffic




                                                                                26
                                      Wavelet De-noising Filter
                                 Applying                                             a threshold-based wavelet filter
                                                           Incoming traffic volume u(t) is decomposed into
                                                            wavelet format
                                                                                                                                                                                   
                                                                                                                           J
                                                             u(t)                          2 t  k   d j k   2 j / 2  2 j t  k
                                                                                    c k  2
                                                                                       j0
                                                                                                 j0 / 2

                                                                                                        j j k
                                                                                                              j0

                                                                                        
                                                                                    
                                                                                   k
                                                                                                    0  
                                                                                                                                      
                                                                                            Baseband Signal                     Signals at high frequencies
                                                                                                          Higher threshold at the periodic cycle of responsive traffic
500


450                                                                                                                                                           500


400
                                                                                                                                                              450

350


300
                                                                                 Wavelet                                                                      400


                                                                                                                                                              350

250


200


150
                                                                                 decomposition                                                                300


                                                                                                                                                              250


                                                                                                                                                              200
100

                                                                                                                                                              150
50


 0                                                                                                                                                            100
      0   1000   2000   3000   4000   5000   6000   7000   8000   9000   10000

                                                                                                                                                              50


                                                                                                                                                               0
                                                                                                                                                                    0   1000   2000   3000   4000   5000   6000   7000   8000   9000   10000




                               u(t)                                                                                                                                     De-noised traffic in
                                                                                                                                                                       which non-responsive
                                             Thresholds T at                                                                                                            traffic changes are
                                             different time scales                                                                                                      removed        27
                                                                                                                   High bursts to remove
          Wavelet De-noising filter
                       
     Threshold vector T is the solution to the
        following two problems
          MSE          problem
AQM packet loss rate at sample period k
                             n 
  De-nosing error                                                               2
                             Tc y c le          i Tc y c le
                                                                              
                min E                              pk   Tcycle  ploss 
                                            k i 1T 1
                                                                ˆ
                                                                              
                               i 1                        c y c le          
                                            Loss rate of De-      Loss rate of estimated
                s.t.                        noised Traffic        responsive traffic
                    i

                 u k   u k   B  Q
                k 1
                            ˆ                           ref   , i  [1...n]               A 
               Cumulative volume of                                Queue size used by      28
                                                 Extra buffer
               de-noised traffic                                   responsive traffic
Wavelet De-noising filter
 Linear     approximation of MSE
                              n
     min E    p k 
                 ˆ
                           i 1

     s.t.
            i Tc y c le

               pk   T
              
                ˆ
     k  i 1 Tc y c le1
                                  cycle    ploss      Linear approximation condition

       i

      u t   u t   B  Q
     k 1
                 ˆ                            ref   , i  [1...n]            B


                                                                                29
      Wavelet De-noising:
      Performance evaluation
       Single   bottleneck scenario: configure i)




50 long-
term TCP
flows
with the
same RTT
                  R0   Router with RED output queue
                                                      30
                                    Wavelet De-noising:
                                    Performance evaluation
                                     Performance                        metrics in configuration i)
                                7
                         x 10
                    10                                                                                             -1
                                                                                                              10
                     9                                                                                                                                        BLUE
                                                                                                                                                              RED
                     8                                                                                                                                        WaveA
TCP Goodput (bps)




                                                                         BLUE




                                                                                           Packet Loss Rate
                                                                         RED                                                                                  Wave
                     7                                                                                                                                               B
                                                                         WaveA                                                                                Wave
                                                                                                                                                                     C
                     6                                                   Wave                                      -2
                                                                                B
                                                                                                              10
                                                                         Wave
                                                                                C
                     5

                     4

                     3

                     2                                                                                             -3
                      0             500   1000        1500        2000   2500       3000                      10
                                                 Buffer Size (KB)                                                   0   500   1000        1500        2000   2500        3000
                                                                                                                                     Buffer Size (KB)


                                    Varying available buffer sizes
                     The proposed scheme achieves a comparable TCP goodput rate
                     to BLUE with much less packet loss rate
                                                                                                                                                                         31
       Wavelet De-noising:
       Performance evaluation
       Multiple-bottleneck        scenario: configuration iii)
25 long-term flows with
uniform distributed RTTs




25 long-term
TCP flows: 10
flows with
60ms RTT and
15 with 180ms      R1      Router with RED output queue
RTT                                                           32
                                           Wavelet De-noising:
                                           Performance evaluation
                                    Performance                     metrics in configuration iii)
                                           Two bottlenecks:       Bottleneck                                                                      8
                                                                                                                                                      x 10
                                                                                                                                                             7




                                           link1 and link2        moves to link2                                                                  7




                                                                                                                       Group TCP1 Goodput (bps)
                                       7
                                x 10
                           11                                                                                                                     6
                                                                                                                                                                                           BLUE
                                                                                                                                                                                           RED
                                                                                                                                                  5                                        WaveA
                                                                              BLUE
                           10                                                                                                                                                              Wave
                                                                              RED
Group TCP2 Goodput (bps)




                                                                                                                                                                                                  B
                                                                                                                                                  4
                                                                                                                                                                                           Wave
                                                                                                                                                                                                  C
                                                                              Wave
                            9                                                        A                                                            3
                                                                              Wave
                                                                                     B                                                            2
                            8                                                 Wave
                                                                                     C                                    1
                                                                                                                           0 108
                                                                                                                            x                                    50                  100                  150
                                                                                                                       1.14
                            7                                                                                                                                    Link1 Bandwidth (Mbps)

                                                                                                                       1.12

                            6




                                                                                               All TCP Goodput (bps)
                                                                                                                                   1.1
                                                                                                                                                                                            BLUE
                                                                                                                       1.08                                                                 RED
                            5                                                                                                                                                               Wave
                                                                                                                                                                                                      A
                                                                                                                       1.06
                                                                                                                                                                                            Wave
                                                                                                                                                                                                      B
                                                                                                                       1.04                                                                 WaveC
                            4
                                                                                                                       1.02

                            3                                                                                                                     1
                             0                      50                  100              150
                                                                                                                       0.98
                                                    Link1 Bandwidth (Mbps)                                                 0                                     50                  100                  150
                                                                                                                                                                 Link1 Bandwidth (Mbps)

                           The proposed scheme achieves the good TCP goodput in multiple
                           bottleneck scenario with two groups of TCP flows with different RTTs
                                                                                           33
        Wavelet De-noising:
        Performance evaluation
        Performance                           metrics in configuration iii)
                                          Two bottlenecks:       Bottleneck
                                     -2
                                          link1 and link2        moves to link2
                                10
             Packet Loss Rate




                                                                                  BLUE
                                     -3                                           RED
                                10                                                Wave
                                                                                         A
                                                                                  Wave
                                                                                         B
                                                                                  WaveC




                                     -4
                                10
                                      0           50                  100                    150
                                                  Link1 Bandwidth (Mbps)
The proposed scheme achieves lower packet loss rate in multiple
bottleneck scenario with two groups of TCP flows with different RTTs34
     Wavelet De-noising:
     Performance evaluation
   Computation cost with higher link-speed




    Computational cost of the proposed scheme remains relatively
    constant with increasing link speed

                                                                   35
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
 Introduction
 Solutions: overview
 Wavelet de-noising scheme in the core
  network
 Smooth handoff scheme at the edge
 Conclusion



                                          36
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Basic ideas
       Sending a packet on multiple paths during handoff reduces
        loss
            When a packet is lost on one path due to handoff, the packet is
             still available on the other paths                 COA1 is registered
                                                                     to Home Agent and
                                                                     Corresponding
                                                                     Node and Path1 is
                                                                     used to send
                                                                     packets to COA1




                                                                                  37
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Basic ideas
       Sending a packet on multiple paths during handoff reduces
        loss
            When a packet is lost on one path due to handoff, the packet is
             still available on the other path
                                                                                                               Path2 to COA2
                              Home Agent   COA1    Life Time 1                                                 and path1 to
                                           COA2    Life Time 2                                                 COA1 are both
                                                                                                               used to multicast
                                                                                                               data packets to
                  Home Agent
                                                  Internet                   Path2 to COA2
                                                                                              Base Station2
                                                                                                               the mobile node




                                                                                             Mo tion
                                                                                                 ving
                                                                                               irec
                                                                                                        COA2




                                                                                             D
                                                                 Wireless Gateway
                                                                                                        COA1
                                                                                              Mobile Node

                                                                     Path1 to COA1


                 Corresponding Node
                   Corresponding Node      COA1    Life Time 1
                                           COA2    Life Time 2


                                                                                                                             38
                                                                                     Base Station1
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Basic ideas
     Adapt       to different amounts of bandwidth
          Perform rate control on multiple paths during the handoff

                                                                                                             Available bandwidth on
                            Home Agent   COA1    Life Time 1                                                 each path is detected and
                                         COA2    Life Time 2                                                 proper transmission rate
                                                                                                             on each path is
                                                                                                             established during the
                Home Agent
                                                Internet                   Path2 to COA2
                                                                                            Base Station2
                                                                                                             handoff




                                                                                           Mo tion
                                                                                               ving
                                                                                             irec
                                                                                                      COA2




                                                                                           D
                                                               Wireless Gateway
                                                                                                      COA1
                                                                                            Mobile Node

                                                                   Path1 to COA1


               Corresponding Node
                 Corresponding Node      COA1    Life Time 1
                                         COA2    Life Time 2


                                                                                                                           39
                                                                                   Base Station1
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
   Basic ideas
     Adapt       to different amounts of bandwidth
          Multi layer video transmission on multiple paths during handoff

                                                                                                             Optimize the usage of
                            Home Agent   COA1    Life Time 1                                                 different bandwidths for
                                         COA2    Life Time 2                                                 improved video quality

                Home Agent
                                                Internet                   Path2 to COA2
                                                                                            Base Station2




                                                                                           Mo tion
                                                                                               ving
                                                                                             irec
                                                                                                      COA2




                                                                                           D
                                                               Wireless Gateway
                                                                                                      COA1
                                                                                            Mobile Node

                                                                   Path1 to COA1


               Corresponding Node
                 Corresponding Node      COA1    Life Time 1
                                         COA2    Life Time 2


                                                                                                                            40
                                                                                   Base Station1
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
   Multi-path transport protocol design




                                           41
            Simulation Settings
          Simulation Scenario:                                             Different Average background
                                                                            traffic volume in different base
       Home Agent
                                                                            stations are explored in simulation
                                             Intermediate Routers

                                                                                         Wireless Gateways




Corresponding Node
(source of video traffic)

                                                                                                  Base Stations


                                                                                              Distance between
                            Mobile Node                                                       neighboring base
                                                                                              stations = 500 meters
                            (receiver of video traffic)

                                                            Base station cell coverage           Background
                                                            radius = 300 meters with             Traffic Nodes
                                                            link speed = 11 Mbps                            42
      Simulation Settings
   Compared handoff schemes:
     Single path schemes with single mobile IP binding:
        No forwarding: no local packet forwarding for mobile nodes is
         performed among base stations
                Basic Mobile IP technique
          Forwarding: local packet forwarding service is enabled by
           fast mobile IP handoff protocol among base stations
                Represent network layer mobility enhancement techniques that
                 repair the packet loss on a broken path for an active session




                                                                            43
                      Simulation Results
                                                                                                                                           when mobile node is in a high
                                                                                                                                           bandwidth cell, high bandwidth
                    Results and observations:                                                                                             is used to transmit base layer
                                                                                                                                           video
                              Video throughput when the mobile node moves
                               from high bandwidth cell to low bandwidth cell
                                                                                                                                           During handoff, base layer
                                                                                                                                           rate reduces to the rate equal
                                                Multi-path Video Throughput
                                                                                                                                           to the bandwidth in the low
                 6000000                                                                                                                   bandwidth cell, and
                 5000000                                                                                                                   enhancement layer video is
Bit Rate (bps)




                 4000000                                                                                                                   transmitted at the rate equal
                 3000000
                                                                                                                                           to the difference between
                 2000000
                 1000000                                                                                                                   high and low bandwidth in the
                       0                                                                                                                   two cells
                           0
                               3
                                   6
                                       9
                                           12
                                                15
                                                      18
                                                           21
                                                                24
                                                                     27
                                                                          30
                                                                               33
                                                                                    36
                                                                                         39
                                                                                              42
                                                                                                   45
                                                                                                        48
                                                                                                             51
                                                                                                                  54
                                                                                                                       57
                                                                                                                            60
                                                                                                                                 63
                                                                                                                                      66
                                                                          Time (sec)                                                       when mobile node is in a low
                                                                                                                                           bandwidth cell, low bandwidth
                                                     Total Video Throughput              Base Layer Throughput
                                                                                                                                           is used to transmit base layer
                                                                                                                                           video
                                                                                                                                                                    44
Simulation Results
   Reduced packet loss               Multi-path handoff scheme keeps the
                                      packet loss ratio low. Base layer is
                                      protected with near-to-zero loss ratio




                 With different available bandwidth in the new cell




                                                                               45
Simulation Results
 Improved    goodput
     With protection of base layer, the goodput is improved in
      terms of smooth video frame rate




                                                                  46
    Congestion Control in the Internet
    with Mixed Traffic Sources and
    Heterogeneous Access Networks
 Introduction
 Solutions: overview
 Wavelet de-noising scheme in the core
  network
 Smooth handoff scheme at the edge
 Conclusion



                                          47
     Congestion Control in the Internet
     with Mixed Traffic Sources and
     Heterogeneous Access Networks
   Conclusion
       Designed a de-nosing framework to improve the congestion control in
        the core network with mixed traffic patterns
            No packet header examination
                  Avoid network administrative and security restrictions on packet header
                   examination
                  Do not rely on protocol header to identify application traffic pattern
            Scalable to the number of traffic flows
                  No flow identification required
            Scalable to increasing line-speed
                  Algorithms run outside the data path in the routers
            Improved performance in multi-bottleneck heterogeneous traffic conditions
                  High throughput and link utilization
                  Reduced packet loss rate




                                                                                             48
Congestion Control in the Internet
with Mixed Traffic Sources and
Heterogeneous Access Networks
 Designed       an end-system smooth handoff scheme
     Allow adaptive congestion control when traversing different
      wireless access networks
           Reduced packet losses during the handoff
           No transmission rate disruption
     Optimize the usage of different available bandwidths in
      different wireless networks for smooth transmission of video
      streaming
           Allow smooth transition of video quality between different
            networks
           Improved video quality during handoff



                                                                         49
         Congestion Control in the Internet
         with Mixed Traffic Sources and
         Heterogeneous Access Networks
   Publication list
        Journal and Book Chapters
               J. Lu, Y. Pan, K. Fujii, et al, “Adaptive Networks”, book chapter in “Cognitive Networks” by Wiley Inc., Sept, 2007
               Y. Pan, M. Lee, J.B. Kim, T. Suda, “An End-to-End Multi-Path Smooth Handoff Scheme for Stream Media”, in the IEEE
                Journal of Selected Areas of Communications (JSAC), Special-Issue on All-IP Wireless Networks, Vol. 22, No. 4, pp. 653-
                663, May 2004
        Conferences and Workshops
               Y. Pan, “Detecting and Filtering Non-responsive Traffic in AQM Queues without Packet Header Examination”, in Proc. of
                ACM SIGMETRICS’08 Student Thesis Panel, June, 2008, Annapolis, MD
               Y. Pan, W. Tsai, and T. Suda, “Applying Wavelet De-noising to Improve TCP Throughput in AQM queues with Existence of
                Unresponsive Traffic,” in Proc. of IEEE ICCCN 2007, Honolulu, HI
               Y. Pan, M. Lee, J.B. Kim, T. Suda, “An End-to-End Multi-Path Smooth Handoff Scheme for Stream Media”(short version), in
                the Proceedings of ACM WMASH’03 Workshop, page 64-74, Sept. 2003, San Diego, CA
               Y. Pan, M. Lee, J. B. Kim, T. Suda, “Smooth Handoff Scheme for Stream Media with Bandwidth Disparity in Wireless Cells”,
                in the Proceedings of IEEE CCW’03 Workshop, page 9-16, Oct. 2003, Dana Point, CA
        Technical Reports
               J. Lu, Y. Pan, S. Yamamoto, and T. Suda, “Robust Data Dissemination for Wireless Sensor Networks in Hostile
                Environments,” Technical Report 08-10, Bren School of Information and Computer Science, University of California at Irvine
               J. Lu, Y. Pan, J. Wang, A. Yahaya, and T. Suda, "A Cross-layer Analysis Model for Wireless Sensor Network QoS“,
                Technical Report 08-08, Bren School of Information and Computer Science, University of California at Irvine
        In submission:
               Y. Pan, W. Tsai, and T. Suda, “Detecting and Filtering nonresponsive Traffic in AQM Queues using Wavelet De-noising
                Techniques”, submitted to IEEE/ACM Transaction on Networks




                                                                                                                                      50
Thanks!




          51