Mechanism Design for Mobile Phone Sensing Dejun Yang, Guoliang - - PowerPoint PPT Presentation

▶
mechanism design for mobile phone sensing
SMART_READER_LITE
LIVE PREVIEW

Mechanism Design for Mobile Phone Sensing Dejun Yang, Guoliang - - PowerPoint PPT Presentation

Mar 26, 2023 299 likes •665 views

Crowdsourcing to Smartphones: Incentive Mechanism Design for Mobile Phone Sensing Dejun Yang, Guoliang (Larry) Xue, Xi Fang and Jian Tang Arizona State University Syracuse University Global Smartphone Users >2B 1.08B 500M 2010 2012

slide-1
SLIDE 1

Crowdsourcing to Smartphones: Incentive Mechanism Design for Mobile Phone Sensing

Dejun Yang, Guoliang (Larry) Xue, Xi Fang and Jian Tang

Arizona State University Syracuse University

slide-2
SLIDE 2

Global Smartphone Users

500M 1.08B >2B 2010 2012 2015

Date Source: IDC http://www.idc.com/getdoc.jsp?containerId=233553, Go-Gulf http://www.go-gulf.com/blog/smartphone Image source: http://www.foxshop.seeon.com/images/smartphone_shadow-group.jpg

2/36

slide-3
SLIDE 3

Mobile Phone Sensing Apps

Image source: http://www.mynewplace.com/blog/files/2011/05/smart-phone-user.jpg, http://serc.carleton.edu/images/sp/library/google_earth/google_maps_new_york.v2.jpg, http://media.treehugger.com/assets/images/2011/10/nextfest-peir-001.jpg

3/36

slide-4
SLIDE 4

What is Missing?

CPU Memory Power

Smartphone users consume their own resource

4/36

slide-5
SLIDE 5

Related Works

  • S. Reddy
  • D. Estrin M.B. Srivastava
  • Developed recruitment frameworks
  • Focused on user selection, not

incentive design

  • Developed a sealed-bid second-price

auction

  • The platform utility was not

considered

  • Designed an auction based dynamic

price incentive mechanism

  • Truthfulness was not considered
  • G. Danezis, S. Lewis, and R. Anderson; “How Much is Location Privacy Worth?” In WEIS 2005.
  • G. Danezis
  • S. Lewis
  • R. Anderson

J-S. Lee and B. Hoh; “Sell Your Experiences: Market Mechanism based Participation Incentive for Participatory Sensing” in PERCOM 2010

J-S. Lee

  • B. Hoh
  • S. Reddy, D. Estrin, and M.B. Srivastava; “Recruitment framework for participatory sensing data collections” in PERVASIVE 2010

5/36

slide-6
SLIDE 6

Other Related Works

Energy Saving MAUI (MobiSys 2010) Application Development PRISM (MobiSys 2010) Privacy PEPSI (WiSec 2011) TP (HotNets 2011)

6/36

slide-7
SLIDE 7

Outline/Progress

Related Works System Model Platform-Centric Model User-Centric Model Simulation Results Conclusions

7/36

slide-8
SLIDE 8

System Model

Platform-Centric Model User-Centric Model

𝑉 = {1, 2, … , 𝑜}, 𝑜 ≥ 2

8/36

slide-9
SLIDE 9

Platform-Centric Model

  • Platform announces a total reward 𝑆
  • Each user 𝑗 has the sensing time 𝑢𝑗 ≥ 0 and sensing

cost 𝜆𝑗 × 𝑢𝑗, where 𝜆𝑗 is its unit cost

  • The utility of user 𝑗 is

𝑣 𝑗 = 𝑢𝑗 𝑢𝑘

𝑘∈𝑉

𝑆 − 𝑢𝑗𝜆𝑗

  • The utility of the platform is

𝑣 0 = 𝜇 log 1 + log 1 + 𝑢𝑗

𝑗∈𝑉

− 𝑆 where 𝜇 > 1 is a system parameter.

9/36

slide-10
SLIDE 10

User-Centric Model

  • Platform announces a set Γ = {𝜐1, 𝜐2, … , 𝜐𝑛} of

tasks, where each 𝜐𝑘 has a (private) value 𝜉

𝑘 > 0.

  • Each user 𝑗 ∈ 𝑉 selects a subset Γ𝑗 ⊆ Γ, based on

which user 𝑗 has a (private) cost 𝑑𝑗

Auction

Γ

1, 𝑐1 , …, Γ𝑜, 𝑐𝑜

𝑇 𝑞1, 𝑞2, … , 𝑞𝑜

  • Utility of user 𝑗 is 𝑣

𝑗 = 𝑞𝑗 − 𝑑𝑗, 𝑗𝑔 𝑗 ∈ 𝑇, 0, 𝑝𝑢ℎ𝑓𝑠𝑥𝑗𝑡𝑓.

  • Utility of the platform is 𝑣

0 = 𝜉 𝑇 − 𝑞𝑗

𝑗∈𝑇

,where 𝜉 𝑇 = 𝜉

𝑘 𝜐𝑘∈∪𝑗∈𝑇Γ𝑗

.

10/36

slide-11
SLIDE 11

Mobile Phone Sensing System

Sensing Task Desc. Sensing Plan Sensed Data Smartphone Users Platform

11/36

slide-12
SLIDE 12

Outline/Progress

Related Works System Model Platform-Centric Model User-Centric Model Simulation Results Conclusions

12/36

slide-13
SLIDE 13

Stackelberg Game (Platform-Centric)

Leader Followers

Stackelberg Equilibrium:

  • Each follower tries to maximize its utility, given

the leader’s strategy

  • The leader tries to maximize its utility, given the

knowledge of the followers’ behavior

13/36

slide-14
SLIDE 14

User Sensing Time Determination

Leader Followers

Sensing Time Determination (STD) game: Players: Users Strategy: Sensing Time Utility: 𝑣 𝑗 =

𝑢𝑗 𝑢𝑘

𝑘∈𝑉

𝑆 − 𝑢𝑗𝜆𝑗

14/36

slide-15
SLIDE 15

NE Computation

Sort users according to their unit costs, 𝜆1 ≤ 𝜆2 ≤ ⋯ ≤ 𝜆𝑜. 𝑇 ← {1, 2}, 𝑗 ← 3; while 𝑗 ≤ 𝑜 𝑏𝑜𝑒 𝜆𝑗 <

𝜆𝑗+ 𝜆𝑘

𝑘∈𝑇

|𝑇|

𝑇 ← 𝑇 ∪ 𝑗 , 𝑗 ← 𝑗 + 1; end for each 𝑗 ∈ 𝑉 if 𝑗 ∈ 𝑇 then 𝑢𝑗

𝑜𝑓 = 𝑇 −1 𝑆 𝜆𝑘

𝑘∈𝑇

1 −

𝑇 −1 𝜆𝑗 𝜆𝑘

𝑘∈𝑇

; else 𝑢𝑗

𝑜𝑓 = 0;

return 𝑢1

𝑜𝑓, 𝑢2 𝑜𝑓, … , 𝑢𝑜 𝑜𝑓

THEOREMs 1&2: The strategy profile 𝑢𝑜𝑓 = 𝑢1

𝑜𝑓, 𝑢2 𝑜𝑓, … , 𝑢𝑜 𝑜𝑓 is

the unique NE of the STD game.

Leader Followers

15/36

slide-16
SLIDE 16

Platform Reward Determination

𝑣 0 = 𝜇 log 1 + log 1 + 𝑌𝑗𝑆

𝑗∈𝑇

− 𝑆 where 𝑌𝑗 =

𝑇 −1 𝜆𝑘

𝑘∈𝑇

1 −

𝑇 −1 𝜆𝑗 𝜆𝑘

𝑘∈𝑇

THEOREM 3: There exists a unique SE R∗, 𝑢𝑜𝑓 in the MSensing game, where 𝑆∗ is the unique maximizer of the above utility function, which is strictly concave.

Leader Followers

𝑣 0 = 𝜇 log 1 + log 1 + 𝑢𝑗

𝑗∈𝑉

− 𝑆

16/36

slide-17
SLIDE 17

Outline/Progress

Related Works System Model Platform-Centric Model User-Centric Model Simulation Results Conclusions

17/36

slide-18
SLIDE 18

LSB Auction (Not Truthful)

𝑇 ← 𝑗 , where 𝑗 ← arg max

i∈𝑉 𝑔

𝑗 ; ⋇while ∃𝑗 ∈ 𝑉 ∖ 𝑇 such that 𝑔 𝑇 ∪ 𝑗 > 1 +

𝜗 𝑜2 𝑔 𝑇

𝑇 ← 𝑇 ∪ {𝑗}; if ∃𝑗 ∈ 𝑇 such that 𝑔 𝑇 ∖ 𝑗 > 1 +

𝜗 𝑜2 𝑔 𝑇

𝑇 ← 𝑇 ∖ {𝑗}; go to ⋇; if 𝑔 𝑉 ∖ 𝑇 > 𝑔 𝑇 then 𝑇 ← 𝑉 ∖ 𝑇; for each 𝑗 ∈ 𝑉 if 𝑗 ∈ 𝑇 then 𝑞𝑗 ← 𝑐𝑗; else 𝑞𝑗 ← 0; return (𝑇, 𝑞)

𝑔 𝑇 = 𝑣 0 𝑇 + 𝑐𝑗

𝑗∈𝑉

is 𝑡𝑣𝑐𝑛𝑝𝑒𝑣𝑚𝑏𝑠 and nonnegative

18/36

slide-19
SLIDE 19

Truthful Auction

THEOREM 5: An auction mechanism is truthful if and only if, for any bidder i and any fixed choice of bid b-i by other bidders, 1)The selection rule is monotonically nondecreasing in 𝑐𝑗; 2)The payment pi for any winning bidder i is set to the critical value.

19/36

slide-20
SLIDE 20

MSensing Auction

Winner Determination Pricing 𝑇 ← ∅, 𝑗 ← arg max

𝑘∈𝑉 𝑤𝑘 𝑇 − 𝑐 𝑘 ;

while 𝑐𝑗 < 𝑤𝑗 and 𝑇 ≠ 𝑉 𝑇 ← 𝑇 ∪ {𝑗}; 𝑗 ← arg max

𝑘∈𝑉∖𝑇 𝑤𝑘 𝑇 − 𝑐 𝑘 ;

20/36

slide-21
SLIDE 21

MSensing Auction

Winner Determination Pricing

𝑞𝑗 ← 0 for all 𝑗 ∈ 𝑉; for each 𝑗 ∈ 𝑇 𝑉′ ← 𝑉 ∖ {𝑗}, 𝑈 ← ∅; repeat 𝑗𝑘 ← arg max

j∈𝑉′∖𝑈(𝑤𝑘 𝑈 − 𝑐 𝑘);

𝑞𝑗 ← max 𝑞𝑗, min 𝑤𝑗 𝑈 − 𝑤𝑗𝑘 𝑈 − 𝑐𝑗𝑘 , 𝑤𝑗 𝑈 ; 𝑈 ← 𝑈 ∪ {𝑗𝑘}; until 𝑐𝑗𝑘 ≥ 𝑤𝑗𝑘 or 𝑈 = 𝑉′; if 𝑐𝑗𝑘 < 𝑤𝑗𝑘 then 𝑞𝑗 ← max 𝑞𝑗, 𝑤𝑗 𝑈 ;

21/36

slide-22
SLIDE 22

Walk-through Example (MSensing)

1 2 3 4 5 8 6 6 3 8 6 8 10 5 6 9

𝑇 = ∅: 𝑤1 ∅ − 𝑐1 = 𝑤 ∅ ∪ 1 − 𝑤 ∅ − 𝑐1 = 19, 𝑤2 ∅ − 𝑐2 = 18, 𝑤3 ∅ − 𝑐2 = 17 𝑤4 ∅ − 𝑐4 = 1. 𝑇 = {1}: 𝑤2 1 − 𝑐2 = 𝑤 1 ∪ {2} − 𝑤 1 − 𝑐2 = 2, 𝑤3 1 − 𝑐3 = 3, 𝑤4 1 − 𝑐4 = −5. 𝑇 = {1,3}: 𝑤2 1,3 − 𝑐2 = 𝑤 1,3 ∪ {2} − 𝑤 1,3 − 𝑐2 = 2, 𝑤4 1 − 𝑐4 = −5. 𝑇 = {1,3,2}: 𝑤4 1,3,2 − 𝑐4 = −5. Winner Selection:

1 2 3 4

22/36

slide-23
SLIDE 23

Walk-through Example (MSensing)

1 2 3 4 5 8 6 6 3 8 6 8 10 5 6 9

𝑞1: Winners are {2,3}. 𝑤1 ∅ − 𝑤2 ∅ − 𝑐2 = 9, 𝑤1 {2} − 𝑤3 2 − 𝑐3 = 0, 𝑤1 2,3 = 3. 𝑞1 = 9 ≥8. Payment Determination: 𝑞2: Winners are {1,3}. 𝑤2 ∅ − 𝑤1 ∅ − 𝑐1 = 5, 𝑤2 {1} − 𝑤3 1 − 𝑐3 = 5, 𝑤2 1,3 = 8. 𝑞2 = 8 ≥6. 𝑞3: Winners are {1,2}. 𝑤3 ∅ − 𝑤1 ∅ − 𝑐1 = 4, 𝑤3 {1} − 𝑤2 1 − 𝑐2 = 7, 𝑤3 1,2 = 9. 𝑞3 = 9 ≥6.

1 2 3 4

23/36

slide-24
SLIDE 24

MSensing is Truthful

THEOREM 6. MSensing is computationally efficient, individually rational, profitable and truthful.

24/36

slide-25
SLIDE 25

Outline/Progress

Related Works System Model Platform-Centric Model User-Centric Model Simulation Results Conclusions

25/36

slide-26
SLIDE 26

Simulation Setup

  • Platform-Centric Model

– 𝑜 is varied from 100 to 1000 – Cost is uniformly distributed over [1, 𝜆𝑛𝑏𝑦], where 𝜆𝑛𝑏𝑦 is varied from 1 to 10 – 𝜇 is set to 3, 5, 10

  • User-Centric Model

– 𝑜 is varied from 1000 to 10000 – 𝑛 is varied from 100 to 500 – 𝜗 is set to 0.01

26/36

slide-27
SLIDE 27

Platform-Centric Incentive Mechanism

Running Time

27/36

slide-28
SLIDE 28

Platform-Centric Incentive Mechanism

Number of Participating Users

28/36

slide-29
SLIDE 29

Platform-Centric Incentive Mechanism

Platform Utility

29/36

slide-30
SLIDE 30

Platform-Centric Incentive Mechanism

User Utility

30/36

slide-31
SLIDE 31

Simulation Setup

  • User-Centric Model

31/36

slide-32
SLIDE 32

User-Centric Incentive Mechanism

Running Time

32/36

slide-33
SLIDE 33

User-Centric Incentive Mechanism

Platform Utility

33/36

slide-34
SLIDE 34

User-Centric Incentive Mechanism

𝑑333 = 3 𝑑851 = 18 Verification of Truthfulness

34/36

slide-35
SLIDE 35

Outline

Related Works System Model Platform-Centric Model User-Centric Model Simulation Results Conclusions

35/36

slide-36
SLIDE 36

Conclusions

Designed incentive mechanisms for mobile phone sensing Platform-Centric Model

  • Modeled as a Stackelberg game
  • Proved the uniqueness of Stackelberg Equilibrium, which can be

computed efficiently

User-Centric Model

  • Modeled as an auction
  • Proved the computational efficiency, individual rationality, profitability

and truthfulness

36/36