Exploiting Mobility in Ad-Hoc Wireless Networks with Incentives

Document Sample
Exploiting Mobility in Ad-Hoc Wireless Networks with Incentives Powered By Docstoc
					Exploiting Mobility in Ad-Hoc
Wireless Networks with
Incentives

       Daniel Figueiredo
        Michele Garetto
         Don Towsley
Outline
   Incentive mechanisms and Mobility
   System model
   Game model
   Simulation results
Incentives
   Incentive Problem:
       Battery power and bandwidth are scarce
        resource
       Refuse to cooperate / relay traffic
   Solutions
       Reputation mechanisms
       Payments / Pricing mechanisms
Mobility
   Nodes are allowed to move strategically.

   Under different incentive mechanisms
       In reputation systems, nodes may want to
        maintain high ―reputations‖.
       In payment mechanisms, nodes may want
        to minimize their costs.
Outline
   Incentive and Mobility
   System model
   Game model
   Simulation results
System Model

        S


            B



            r




                D
System Model – relayed traffic
intensity R(x)

                      S



                  B

              o   P       x
System Model
   Parameters:
       B rectangular width
       r transmission range
   Metrics
       R(x) relayed traffic intensity
       H(x) average path length
           H(x) =  D Dist(D,S(x)) /r dD
System Model – relationship
between R(x) and H(x)
    xR(x) dx =  xH(x) dx
       Represents the total amount of traffic
        exchanged in the network.
       Two distribution have the same mean.
   Payment mech.  max R(x)-H(x)
   Reputation mech.  min R(x)
Outline
   Incentive and Mobility
   System model
   Game model
   Simulation results
Game Model
   Each node is a player.
   Players take turns to move.
   Assume at any time each node knows
    the location of all other nodes.
Game Model
   Let pi=(xi ,yi) be the position of node i.
   P = (p1,p2,…,pN)
   P-i is P with i-th component removed.
   Redefine:
       R(pi,P-i) amount of relayed traffic by node i.
       H(pi,P-i) average # of hops from i to all
        destinations
Game Model– I(pi,P-i)
   Let I(pi,P-i) be the noise produced by
    other nodes’ transmissions.
   Assume signal power decay distance x
    as 1/xa,with 2<a<4.
   I(pi,P-i) =jiN(j,i) / dist(pi,pj)a
   N(j,i) is the average # of transmissions
    at j that interfere with the reception of a
    useful signal at i.
Game Model—payoff function

   U(pi,P-i)= R(pi,P-i)+H(pi,P-i)+I(pi,P-i)
   For payment mech. =1, =-1
   For reputation mech. =-1, =0
    is negligible.
Subgame perfect equilibrium
& backward induction
                     1
               L          R

           2                    2
      l        r          l         r

   (0,0)           (3,2)(1,4)           (2,1)
Subgame perfect equilibrium
& backward induction
                     1
               L          R

           2                    2
      l        r          l         r

   (0,0)           (3,2)(1,4)           (2,1)
Subgame perfect equilibrium
& backward induction
                     1
               L          R

           2                    2
      l        r          l         r

   (0,0)           (3,2)(1,4)           (2,1)
Outline
   Incentive and Mobility
   System model
   Game model
   Simulation results
Key Observations
   Under reputation mech., nodes tends to
    cluster together.
   Under payment mech., nodes tends to
    scatter.
   By considering the interference, nodes
    keep certain distance with each other.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:12/28/2011
language:
pages:19