Docstoc

Routing in Sensor Networks Directed Diffusion and other proposals

Document Sample
Routing in Sensor Networks Directed Diffusion and other proposals Powered By Docstoc
					        Wireless Sensor Networking
(Understanding the radio, MAC, & routing protocols)



              Romit Roy Choudhury




                                                      1
Sensor Networking – Why ??
  Data Collection – A basic need
     Will the volcano erupt? Need temperature/gas signatures
     How much Global Warming? Need ocean current data
     How many enemy tanks crossed?

  Human monitoring possible/feasible ?
     Often risky, impenetrable, costly, …

  But science has collected data for centuries …
     Manual (wired) placement, periodic human visits
     Wireless data transmitters
     Community accepted barriers/defiiencies



                                                                2
New Opportunities
 Device miniaturization
    Moore’s law
    Processors envisioned as smart dust


 Innovations in wireless communication
    Low power communication
    Antenna sizes smaller with high frequency

      Device + RF + sensors - A new breakthrough:
  Scattered sensor motes self-organize themselves forming a network.
 Sensed data aggregated, processed, and transported to base station.
              Low risk, low cost, and heavy penetration

                                                                       3
Plethora of Applications




                           4
Plethora of Challenges
 Devices
   Reducing energy consumption
   Heavy programming constraints (16 KB RAM)


 Wireless Radio Network
     Reliable low power communication
     Medium access control (MAC)
     Network wide energy conservation
     Routing
     Aggregation, compression, suppression
     …
                                                5
Today’s Talk
 Understanding the wireless channel
   The departure from wireline
   The key challenges


 Medium access control
   Protocol design
   Energy-awareness (coordinated sleeping)


 Routing
   Unicast (Diffusion)
   Broadcast (Gossip)
                                              6
The Wireless Channel




                       7
Many Motivations for Wireless
 Unrestricted mobility / deployability
    Unplugged from power outlet


 Significantly lower cost
    No cable layout, service provision
    Low maintenance


 Ease
    Direct communication with minimum infratructure



                                                       8
From Links to Networks
 Variety of architectures

   Single hop networks




   Multi-hop networks




                             9
The Wireless Future …

                        Internet




                                   10
No Free Lunch
 Numerous challenges
     Channel fluctuation
     Lower bandwidth
     Limited Battery power
     Disconnection due to mobility
     Security
     …




                                      11
Question Is …




  Can’t we use the rich “wireline” knowledge ?
           In solving the wireless challenges




                                                 12
The Answer



    Wireless channel: A dispersive medium
    The PHY and MAC layer completely dissimilar


           The whole game changes




                                                  13
On Our Agenda
 Quick Glimpse
   Medium Access Control
     • Wired
     • Wireless

   The emergence of 802.11

   Evolution of sensor network MAC protocols
     • Energy awareness




                                                14
Medium Access Control




                        15
The Channel Access Problem
 Multiple nodes share a channel

     A                   B                    C




 Pairwise communication desired
   Simultaneous communication not possible

 MAC Protocols
   Suggests a scheme to schedule communication
     • Maximize number of communications
     • Ensure fairness among all transmitters
                                                  16
The Trivial Solution
      A                      B                     C




 Transmit and pray
   Plenty of collisions --> poor throughput at high load




                                                            17
The Simple Fix                         Don’t
                                     transmit

      A                      B                         C




 Transmit and pray
   Plenty of collisions --> poor throughput at high load


 Listen before you talk
   Carrier sense multiple access (CSMA)
   Defer transmission when signal on channel
                                      Can collisions still occur?

                                                                    18
 CSMA collisions
                                 spatial layout of nodes


Collisions can still occur:
Propagation delay non-zero
between transmitters



When collision:
Entire packet transmission
time wasted


note:
Role of distance & propagation
delay in determining collision
probability
                                                           19
CSMA/CD (Collision Detection)

 Keep listening to channel
   While transmitting




 If (Transmitted_Signal != Sensed_Signal)
             Sender knows it’s a Collision
             ABORT
                                              20
2 Observations on CSMA/CD


 Transmitter can send/listen concurrently
   If (Sensed - received = null)? Then success


 The signal is identical at Tx and Rx
   Non-dispersive




       The transmitter can DETECT if and
              when collision occurs
                                                  21
Unfortunately …




   Both observations do not hold for wireless




                                  Leading to …

                                                 22
Wireless Medium Access Control

                             C          D
              A       B


         Signal
         power




  SINR threhold




                             Distance

                                            23
Wireless Media Disperse Energy
A cannot send and listen in parallel
                                                      C              D
                   A                     B


              Signal
              power


                                       Signal not same at different locations


     SINR threhold




                                                      Distance

                                                                                24
Collision Detection Difficult




  Signal reception based on SINR
    Transmitter can only hear itself
    Cannot determine signal quality at receiver




                                                   25
Calculating SINR

                               B
              A                          C




              SignalOfIn  terest ( SoI )
     SINR 
            Interference( I )  Noise( N )
                     A
                  Ptransmit                    A
                                             Ptransmit
     SoI A

                   d                   d
          B
                    AB
                              SINRB 
                                  A       AB
                                           C
               C                         Ptransmit
      C P
     I      transmit                N 
      B                                    d CB
         d CB
                                                         26
Red signal >> Blue signal       Red < Blue = collision



                                         C               D
    X
                 A          B


           Signal
           power




  SINR threhold




                                         Distance

                                                             27
Important: C has not heard A, but can interfere at receiver B

                                        C is the hidden terminal to A

                                                 C              D
  X
             A                      B


        Signal
        power




SINR threhold




                                                Distance

                                                                        28
    Important: X has heard A, but should not defer transmission to Y

Y                    X is the exposed terminal to A

                                                      C            D
      X
                 A                        B


            Signal
            power




    SINR threhold




                                                      Distance

                                                                       29
A Project Idea!

                            C          D
        X
                   A    B


              Signal
              power




     SINR threhold

Sensitivity threshold

                            Distance

                                           30
WillProject the wireless MAC problem?
A this solve Idea!                            Do not
                                            transmit in
                                            this region

                                        C              D
        X
                   A             B


              Signal
              power




     SINR threhold
                                                   T
Sensitivity threshold


                                        Distance

                                                           31
The Emergence of 802.11
 Wireless MAC proved to be non-trivial

 1992 - research by Karn (MACA)
 1994 - research by Bhargavan (MACAW)

 Led to IEEE 802.11 committee
   The standard was ratified in 1999




                                          32
IEEE 802.11 with Omni Antenna

 RTS = Request                         CTS = Clear
   To Send                               To Send
                     M
                                       Y

                 S       RTS   D
                         CTS


                                   K




                                                     33
IEEE 802.11 with Omni Antenna


                    silenced
                      M
                                      Y

                    S               silenced
                        Data   D
                         ACK

         X                          silenced
                                   K
         silenced



                                               34
But is that enough?




                      35
RTS/CTS
 Does it solve hidden terminals ?
     Assuming carrier sensing zone = communication zone



                                            E     RTS
                                                          F
                                     CTS

       A        B          C          D




E does not receive CTS successfully  Can later initiate transmission to D.
                    Hidden terminal problem remains.

                                                                              36
Hidden Terminal Problem
 How about increasing carrier sense range ??
   E will defer on sensing carrier  no collision !!!



                                      E   RTS
                                                F
                                CTS

    A       B        C          D
                         Data




                                                         37
Hidden Terminal Problem
 But what if barriers/obstructions ??
   E doesn’t hear C  Carrier sensing does not help



                                    E   RTS
                                              F
                              CTS

     A     B       C          D
                       Data




                                                       38
Exposed Terminal

 B should be able to transmit to A
   RTS prevents this



                               E
               RTS
                         CTS

     A     B         C    D




                                      39
Exposed Terminal

 B should be able to transmit to A
   Carrier sensing makes the situation worse



                                 E
               RTS
                           CTS

     A     B         C      D




                                                40
Thoughts !
 802.11 does not solve HT/ET completely
   Only alleviates the problem through RTS/CTS and
    recommends larger CS zone

 Large CS zone aggravates exposed terminals
   Spatial reuse reduces  A tradeoff
   RTS/CTS packets also consume bandwidth
   Moreover, backing off mechanism is also wasteful

 The search for the best MAC protocol is still on.
     However, 802.11 is being optimized too.
     Thus, wireless MAC research still alive
                                                       41
Questions?




             42
Energy-Awareness
 802.11 optimizes for throughput/latency
   Energy savings is second priority


 Unattended sensor networks
   Operate on AA batteries
   Yet, expected to last for months or years


 Energy-awareness is the key
   Throughput and latency is secondary



                                                43
An Energy-Efficient MAC Protocol for Wireless
        Sensor Networks (S-MAC)



      Wei Ye, John Heidemann, Deborah Estrin




                                                44
Major source of energy waste
 Collision

 Overhearing

 Control Overhead

 Idle Listening
   Listening to possible traffic that is not sent
   50%-100% energy drain compared with receiving




                                                     45
Avenues to Reduce Energy Consumption

(1) Periodic listen and sleep
(2) Collision avoidance
(3) Overhearing avoidance
(4) Message passing




                                       46
(1) Periodic Listen and Sleep

 The main idea
   Put nodes to sleep periodically




   Called “Duty Cycles”

   However, ensure that sleep/wake-up is synchronous



                                                        47
Listen/Sleep Schedule Assignment

Choosing Schedule (1)

                      Listen
                                                                    Synchronizer
                                                                    • Listen for a mount of time
A   Listen for SYNC        Go to sleep after time t         Sleep
                                                                    • If hear no SYNC, select its
                                                                      own SYNC
                                                                    • Broadcasts its SYNC
       Broadcasts
                                                                      immediately


                                                                    Follower
                             Listen                                 • Listen for a mount of time
B                     td     Go to sleep after time t- td   Sleep   • Hear SYNC from A, follow
                                                                       A’s SYNC
                                                                    • Rebroadcasts SYNC after
             Broadcasts
                                                                       random delay td


                                                                                                    48
    Listen/Sleep Schedule Assignment

    Choosing Schedule (2)
                                                                           1. B receives A’s schedule and
                        Listen
                                                                              rebroadcast it.
A                                                              Sleep
      Listen for SYNC        Go to sleep after time t1                     2. Hear different SYNC from C
        Broadcasts                                                         3. Adapt both schedules
                              Listen

B                       td                                     Sleep

          Broadcasts

                                     Only need to broadcast once


                                                                             Nodes only rarely adopt
                                 Listen                                      multiple schedules
C              Listen for SYNC     Go to sleep after time t2       Sleep




                                                                                                            49
Keeping Clocks in SYNC
 SYNC packets must not collide
   Reserve separate time window for SYNC transmission




                                                         50
(2) Collision Avoidance
 Identical to 802.11
   RTS/CTS
   Virtual carrier sense (NAV)
   Physical carrier sense




                                  51
(3) Overhearing Avoidance
Neighbors go to sleepon overhearing RTS/CTS




   A is talking to B
   D receives CTS from B -> sleep
   • D’s transmission will collide B’s
   C receives RTS from A -> sleep
   • C cannot receive CTS/DATA from E

   All immediate neighbours of transmitting node sleep

   How long should they sleep?
   C and D update their NAV
   Keeping sleeping until NAV count down to zero
                                                         52
(4) Message Passing
 How to transmit long message?
         Transmitting one long message is inefficient
         Many small packets with RTS/CTS/ACK for each



 S-MAC: Divide into fragments, transmit in burst
         RTS/CTS reserve medium for the entire sequence
         Fragment-errors recovery with ACK
      •    no control packets for fragments




                                                           53
Acknowledgment to Pro. Jun Yang



                     Neighbors can sleep for whole message




                                                             54
Message Passing

Advantages:
 Energy saving:
    Neighbors go to sleep when sense transmissions
    Reduces control overhead by sending multiple ACK



Disadvantage:
 Node-to-node fairness reduces
   However, message-level latency reduces




                                                        55
             Listen time: 300ms
Experiment   Sleeping time: 1s
             SYNC: every 13s (10 listen/sleep period)
             A, B, C use the same schedule




                                                        56
                         Energy save due
                         to periodic sleep   802.11

Energy save due to avoiding
overhearing by using message
passing                                      OA




                                             SMAC




Heavy Traffic           Light Traffic
                                                      57
OA: In light traffic status, nodes keep
listening for quite a long time




                                          58
SYNC overhead



Overhearing avoidance still benefit




  Heavy Traffic              Light Traffic




                                             59
Questions?




             60
       Studied MAC protocols till now

       Another important challenge:
How does a packet get transported end to end

      i.e., How do you perform Routing




                                               61
The Problem
 A region requires event-
  monitoring (harmful gas,                              A sensor field
  vehicle motion, seismic vibration,
  temperature, etc.)                   Sensor sources

                                                          Event
 Deploy sensors forming a
  distributed network

                                                             Directed
 On event, sensed and/or                                    Diffusion
  processed information
  delivered to the inquiring
  destination                                    Sensor sink




                                                                         62
Directed Diffusion




                     63
The Proposal
 Proposes an application-aware paradigm to
  facilitate efficient aggregation, and delivery of
  sensed data to inquiring destination

 Challenges:
      Scalability
      Energy efficiency
      Robustness / Fault tolerance in outdoor areas
      Efficient routing (multiple source destination pairs)



                                                               64
IP or not to IP
 IP is the pivot of wired/wireless networks
   All networking protocol over and below IP


 Should we stick to this model?

                  Comments ?




                                                65
Directed Diffusion
 Typical IP based networks
   Requires unique host ID addressing
   Application is end-to-end, routers unaware


 Directed diffusion – uses publish/subscribe
   Inquirer expresses an interest, I, using attribute
    values
   Sensor sources that can service I, reply with data




                                                         66
Data Naming
 Expressing an Interest
   Using attribute-value pairs
   E.g., Type = Wheeled vehicle // detect vehicle location
            Interval = 20 ms           // send events every 20ms
            Duration = 10 s            // Send for next 10 s
            Field = [x1, y1, x2, y2]   // from sensors in this area


 Other interest-expressing schemes possible
   E.g., hierarchical (different problem)




                                                                      67
Gradient Set Up

  Inquirer (sink) broadcasts exploratory interest, i1
     Intended to discover routes between source and sink


  Neighbors update interest-cache and forwards i1

  Gradient for i1 set up to upstream neighbor
     No source routes
     Gradient – a weighted reverse link
     Low gradient  Few packets per unit time needed




                                                            68
Exploratory Gradient

                         Exploratory Request
                         Gradient


                          Event


                   Low                    Low
             Low




 Bidirectional gradients established on all links through flooding


                                                                     69
Event-data propagation
 Event e1 occurs, matches i1 in sensor cache
   e1 identified based on waveform pattern matching


 Interest reply diffused down gradient (unicast)
   Diffusion initially exploratory (low packet-rate)


 Cache filters suppress previously seen data
   Problem of bidirectional gradient avoided




                                                        70
Reinforcement
                              Reinforced gradient
                                      Reinforced gradient
                 Event
        A sensor field                    Sink

 From exploratory gradients, reinforce optimal
  path for high-rate data download  Unicast

   By requesting higher-rate-i1 on the optimal path

   Exploratory gradients still exist – useful for faults



                                                            71
Path Failure / Recovery
 Link failure detected by reduced rate, data loss
   Choose next best link (i.e., compare links based on
    infrequent exploratory downloads)
 Negatively reinforce lossy link
   Either send i1 with base (exploratory) data rate
   Or, allow neighbor’s cache to expire over time

                                             Link A-M lossy
      Event         D                        A reinforces B
              Src
                           M                 B reinforces C …
                                  A          D need not
                    C                        A (–) reinforces M
                           B          Sink
                                             M (–) reinforces D

                                                                  72
Loop Elimination

               P                   Q



         D               M               A


 M gets same data from both D and P, but P
  always delivers late due to looping
    M negatively-reinforces (nr) P, P nr Q, Q nr M
    Loop {M  Q  P} eliminated
 Conservative nr useful for fault resilience


                                                      73
Simulation Setup & Metrics
 ns2, 50 nodes in 160x160 sqm., range 40m
    Node density maintained, 802.11 MAC
 Random 5 sources in 70x70, random 5 sinks
 Average Dissipated Energy
    Per node energy dissipation / # events seen by sinks
 Average Delay
    Latency of event transmission to reception at sink
 Distinct event delivery ratio
    Ratio of # events sent to # events received by sink



                                                            74
Average Dissipated Energy



                                          flooding
                 Multicast                 Diffusion




 In-network aggregation reduces DD redundancy
   Flooding poor because of multiple paths from source to sink

                                                                  75
Delay




                                              flooding

                                             Diffusion
                    Multicast



 DD finds least delay paths, as OM – encouraging
   Flooding incurs latency due to high MAC contention, collision

                                                                    76
Event Delivery Ratio under node failures

                                                      0%




                                                   10%
                                           20%




  Delivery ratio degrades with higher % node failures
   Graceful degradation indicates efficient negative reinforcement

                                                                      77
Conclusion

  Directed diffusion, a paradigm proposed for event
   monitoring sensor networks
  Energy efficiency achievable
  Diffusion mechanism resilient to fault tolerance
     Conservative negative reinforcements proves useful


  A careful MAC protocol, designed for such specifics,
   can yield further performance gains




                                                           78
            Rumor Routing
               LEACH
                SPIN

Some other proposals for sensor routing



                                          79
Rumor Routing




                80
LEACH
 Proposes clustering of sensors + cluster leaders
   Can aggregate data in single (local) cluster
   Rotating cluster head balances energy consumption
   Cluster formation distributed and energy efficient

      Cluster-head
      always awake




  Member nodes can
  sleep when not Txing


                                                         81
LEACH – The Protocol

  Time is divided into rounds
  A node self-elects itself as the cluster head
       Higher residual energy, higher probability to be head
       Close-by sensors join this cluster-head
       Cluster head does TDMA scheduling and gathers data
       Gathered data compressed based on spatial correlation
       Transmits data to Base Station (@ higher power)
  In the next round, another cluster head elected
     Probabilistic load balancing
     Network lifetime can increase manifolds




                                                                82
SPIN: Information Via Negotiation
 Flooding  many sensors transmit same data
   Redundant
 Make sensors disseminate spatially/temporally
  disjoint data sets
   Name data with meta-data to define space/time
    property
   Sensors compare overheard data with self-sensed
    data
   Combine data to minimize overlap
 Make sensors resource-adaptive
   When low battery  perform minimum activities

                                                      83
The SPIN 3-Step Protocol



                           A


                 B




                               84
The SPIN 3-Step Protocol



                                               A


                       B




    Notice the color of the data packets sent by node B
                                                          85
The SPIN 3-Step Protocol



                                           A


                      B




      SPIN effective when DATA sizes are large :
         REQ, ADV overhead gets amortized
                                                   86
 SmartGossip: A Reliable Broadcast
Service for Wireless Sensor Networks


           Romit Roy Choudhury
       Dept. of ECE, Duke University


              Joint work with
         Pradeep Kyasanur (Google)
           Indranil Gupta (UIUC)




                                       87
Problem
 Broadcast in Sensor Networks
   A widely used service

   Network layer functions heavily depend on it

   Examples:
     • Directed Diffusion
     • Unicast or multicast routing
     • Instruction / code dissemination
     • Query propagation



                                                   88
Approaches
 Several approaches evolved over time



                 Complex
     Broadcast              Single point    Load
                   algo
       Storm                of failures    Balance




                                                     89
Recent Past
 Gossiping = Probabilistic flooding
   Nodes forward with probability, p




Source




                                        90
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p



           Tails
Source




          Heads




                                        91
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p



           Tails
Source




          Heads




                                        92
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p



           Tails
Source




                   Heads
          Heads




                                        93
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p



           Tails
Source




                   Heads
          Heads




                                        94
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p


                   Tails
           Tails
Source
                           Heads



                   Heads
          Heads

                            Tails



                                        95
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p


                   Tails
           Tails
Source
                           Heads



                   Heads
          Heads

                            Tails



                                        96
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p


                   Tails            Tails
           Tails
Source
                           Heads



                   Heads
          Heads                     Heads
                            Tails



                                            97
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p


                   Tails            Tails
           Tails
Source
                           Heads



                   Heads
          Heads                     Heads
                            Tails



                                            98
Recent Past
 Gossip based broadcast
   Nodes forward with probability, p


                   Tails            Tails
           Tails
Source
                           Heads


                                                    Heads
                   Heads
          Heads                     Heads
                            Tails
                                            Tails

                                                            99
 Recent Past
  Gossip based broadcast
      Nodes forward with probability, p


                          Tails            Tails
                Tails
 Source
                                  Heads


                                                           Heads
                          Heads
               Heads                       Heads
                                   Tails
    For carefully chosen ‘p’
 1. Simple,
theFault tolerant                                  Tails
 2. message reaches all nodes
    in minimal transmissions
 3. Load-balanced
                                                                   100
 We Ask …
  Given some topology deployment
      How do we choose a suitable value of “p” ?




  But, for is to simulate
One option this example,
 simulation result will be
  the gossip offline,
            p=1
   and determine “p”
                                                    101
We Ask …
 Given some topology deployment
   How do we choose a suitable value of “p” ?


 Even if topology is homogeneous
   It may change over time due to failure and mobility




                                                          102
We Ask …
 Given some topology deployment
   How do we choose a suitable value of “p” ?


 Even if topology is homogeneous
   It may change over time due to failure and mobility




 Say computed p = 0.85
                                                          103
We Ask …
 Given some topology deployment
   How do we choose a suitable value of “p” ?


 Even if topology is homogeneous
   It may change over time due to failure and mobility
                             Fails




 Say computed p = 0.85
                                                          104
We Ask …
 Given some topology deployment
   How do we choose a suitable value of “p” ?


 Even if topology is homogeneous
   It may change over time due to failure and mobility




                                             15% of packets
                                              will not reach
                                               these nodes
 Say computed p = 0.85
                                                               105
We Ask …
 Given some topology deployment
    How do we choose a suitable value of “p” ?


 Even if topology is homogeneous
    It may change over time due to failure and mobility


 Finally, what if topology is not known a priori ?
    How can you choose “p” ?




                                                           106
We Argue …



            A broadcast service necessary
   that customizes itself to the underlying topology
                            and
         repairs itself with failures and mobility




                                                       107
Smart Gossip
 Intuition
    Identify which of YOUR friends get to know gossips
     earlier than you do
       • Request those friends to gossip more

    Friends who get to know gossips later than you will
     request you to gossip more

    You choose your gossip probability as:
      • MAX value of all requests from YOUR friends



                                                           108
For Example …
 When H spreads a gossip
   F gets gossip only from G
   F asks G to always gossip
   Thus, pG = 1.0

   B receives gossip from A,C,D,E,F
   B also observes that A,C,D,E received gossip from F
     • Indicates that B must depend only on F (parent)
     • A,C,D,E and B are independent (siblings)

   B asks F to always gossip, thus pF = 1.0

                                                          109
For Example …
   B asks F to always gossip,
    thus pF = 1.0

   B does not require siblings
    A,C,D,E to gossip at all

   Thus pA = 0, pC = 0, pD = 0, pE = 0


         Observe that only 2 transmissions
     (from G and F) are sufficient for broadcast

                                                   110
Protocol Details
 For first gossip pkt, nodes transmit with p=1
     Enables nodes to deduce neighbor dependences

     Transmitters piggyback pkt with parent-id from which
      it received the pkt

     Nodes record transmitter-id, and its parent-id, and
      deduce parent, child, sibling relationships …

                                             … see next slide




                                                                111
Deducing Relationships



               SA        Parent = {A}
     S         A             B          C
 Child = {A}

                             E
                         Parent = {A}




                                            112
Deducing Relationships



                             Parent = {A}
     S               A           B             C
 Child = {A}                   AB            Parent = {B}
               Child = {B}

                                 E
                             Parent = {A}
                             Sibling = {B}




                                                        113
Deducing Relationships



                             Parent = {A}
     S               A           B             C
 Child = {A}                                 Parent = {B}
               Child = {B}

                             AE E
                             Parent = {A}
                             Sibling = {B}




                                                        114
Deducing Relationships



                               Sibling = {E}
                               Parent = {A}
     S               A               B           C
 Child = {A}                                   Parent = {B,E}
               Child = {B,E}

                                    E
                               Parent = {A}
                               Sibling = {B}




                                                          115
Choosing Probabilities
 Each node calculates number of parents ( k )
   Assume 99% assurance necessary for gossip
 Node suggests each parent to gossip using ‘p’:
                   0.99 = ( 1 – (1 - p)k )
 Each node receives multiple requests of ‘p’
   Uses Max { pi } as its own gossip probability



    S               A                  B             C
                                                    Parent={B,E}

                                       E

                                                              116
Choosing Probabilities
 Each node calculates number of parents ( k )
   Assume 99% assurance necessary for gossip
 Node suggests each parent to gossip using ‘p’:
                       0.99 = ( 1 – (1 - p)k )
 Each node receives multiple requests of ‘p’
   Uses Max { pi } as its own gossip probability


             p = 1.0                p = 1.0       p = 0.9
    S                   A                     B               C

                                                    p = 0.9
                                p = 1.0
                                              E

                                                                  117
Choosing Probabilities
 Each node calculates number of parents ( k )
   Assume 99% assurance necessary for gossip
 Node suggests each parent to gossip using ‘p’:
                   0.99 = ( 1 – (1 - p)k )
 Each node receives multiple requests of ‘p’
   Uses Max { pi } as its own gossip probability

     p = 1.0        p = 1.0            p = 0.9       p=0
    S               A                  B             C



                                       E
                                           p = 0.9
                                                           118
The Bigger Picture



           Src




                     119
Reliability (Details in paper)
 Node Failures
   Node failures affect broadcast
   Source node flags packet periodically (p=1)
   Allows for updating dependences



 Link Losses
   Node requests upstream nodes to retransmit
     • We require each node to buffer few packets
   Children overhear this request
   Children do not request retransmissions themselves

                                                         120
Evaluation
 Qualnet Simulator, ver 3.7
  (Currently implementing on Moteiv tmotes + TinyOS)


 Metrics used
   Average Reception Percentage
   Average Forwarding Percentage
   Resilience to link/node failures




                                                       121
Percolation



              Smart Gossip



              Adaptive Overhead




              Adaptive Neighbor




                                  122
Forwarding Overhead




                      123
Adaption to Node Failures




     Nodes gossip more to compensate for other failing nodes


                                                               124
Conclusion
 Broadcast is an important problem
   Gossip is good – but not practical for sensor nets
   Need to adapt gossip based on topology / failures


 Smart Gossip
   Form dependence graphs using distributed protocol
   Dependence relations suggest suitable probability


 Results
   Overheads are low, and yet good percolation
   Robust to node and link failures
                                                         125
Questions?




             126
Percolation


                  Smart Gossip



              Adaptive Overhead




              Adaptive Neighbor




                                  127
Wireless Routing
 Link instability causes many routing issues
     Shortest hop routing often worst choice
     Scarce bandwidth makes overhead conspicuous
     Battery power a concern
     Security and misbehavior …


 If that’s not bad enough
   Add node mobility
     • Note: Routes may break, and reconnect later



                                                     128
 Routing in wireless Mobile Networks
  Imagine hundreds of hosts moving
      Routing algorithm needs to cope up with varying
       wireless channel and node mobility
Where’s
RED guy




                                                         129

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:2/8/2012
language:
pages:129