INDAPSON: An Incentive Data Plan Sharing System Based on Self-Organizing Network

1
INDAPSON An Incentive Data Plan Sharing System
Based on Self-Organizing Network
2
Outline

• Introduction
• Motivations
• Objectives
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

2
3
Motivation

• Using mobile terminals to get Internet services
  is becoming increasingly important in peoples
  daily life.
• There exist drawbacks of todays mobile data
  service

• Many ISPs dont provide unlimited data plans.
  their network usage is always dynamic while their
  data plan is fixed. Sometimes their data plans
  are overused but underused at other times.
• The downstream rate offered by current cellular
  networks is insufficient to meet users demand.
  Furthermore, the data rate of cellular network
  fluctuates severely over time.

4
Motivation

• Previous works
• Users interested in the same video can exchange
  the data downloaded from cellular networks1.

1 L. Keller, A. Le, B. Cici, H. Seferoglu and
C. Fragouli, MicroCast cooperative video
streaming on smartphones, in Proc. of
MobiSys12, pp. 57-70, 2012.
5
Motivation

• Previous works
• Users interested in the same video can exchange
  the data downloaded from cellular networks1.
• Mobile terminals connected to the same wireless
  access point can exchange the received packets
  among each other via Bluetooth2.

2 M. Ramadan, L. El Zein and Z. Dawy,
Implementation and evaluation of cooperative
video streaming for mobile devices, in Proc. of
IEEE PIMRC08, pp. 1-5, 15-18, Sept. 2008.
6
Motivation

• Previous works
• Users interested in the same video can exchange
  the data downloaded from cellular networks1.
• Mobile terminals connected to the same wireless
  access point can exchange the received packets
  among each other via Bluetooth2.
• Nodes in close vicinity use the WLAN to discover
  each other, form a collaboration group, and
  stripe traffic across their WWAN links3.

3 G. Ananthanarayanan, V.N. Padmanabhan, L.
Ravindranath and C.A. Thekkath, COMBINE
leveraging the power of wireless peers through
collaborative downloading, in Proc. of
MobiSys07, pp. 286-298, 2007.
7
Motivation

• There exist limitations in the previous works
• All the users must be interested in the same
  video, which implies that a user cannot be
  benefited if the video is cool.
• All the participants must be aware of each other,
  which limits the application scenarios.
• The departure of participants will break off the
  cooperation process, which would be a common
  phenomenon in mobile networks.
• Build a data plan sharing system for mobile
  networks that allows people to exchange their
  sufficient data plans.

8
Objective

• We study
• How to avoid data traffic waste due to fixed data
  plans?
• How to incent people to join the cooperative
  process?
• How to adapt the system to the peculiarity of
  mobile networks?
• We obtain
• A data plan sharing system to allow people
  exchange data plans.
• An incentive mechanism to encourage
  participation.
• A dynamic self-organization strategy to construct
  local wireless networks.

9
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

9
10
System Architecture

• Two layers of networks
• Bluetooth-Management Network (BMN) discovers
  potential participants and manages the
  construction of local download group.
• WiFi-Transmission Network (WTN) exchanges
  downloaded data in high speed.

11
System Architecture

• Three types of users
• Primary user user who needs to download data
  from the Internet. (User A)
• Assistant user user who helps a download user
  download data from the Internet. (User B C)
• Detected user user who acts as a candidate for
  assistant user. (Other users)

12
System Architecture

• Servers
• Management server applies the Reputation
  Adaptive Pricing (RAP) and manages the virtual
  credits.
• Resource server can be any resource servers
  allowing multithreading downloading online.

13
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Self-Organization Scheme
• Cooperative Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

13
14
Self-Organization Scheme

• The query of credits User A gets the size of the
  target data and connects to the management server
  to check whether its surplus credits is enough to
  pay the download.
• The construction of BMN User A (PU) scans
  neighboring users via Bluetooth and invites them
  to be DUs in the BMN.
• The construction of WTN User A initializes its
  soft AP and informs parts of the DUs via
  Bluetooth. The DUs selected connect to the soft
  AP via WiFi.

15
Self-Organization Scheme

• The query of credits User A gets the size of the
  target data and connects to the management server
  to check whether its surplus credits is enough to
  pay the download.
• The construction of BMN User A (PU) scans
  neighboring users via Bluetooth and invites them
  to be DUs in the BMN.
• The construction of WTN User A initializes its
  soft AP and informs parts of the DUs via
  Bluetooth. The DUs selected connect to the soft
  AP via WiFi.

16
Self-Organization Scheme

• The query of credits User A gets the size of the
  target data and connects to the management server
  to check whether its surplus credits is enough to
  pay the download.
• The construction of BMN User A (PU) scans
  neighboring users via Bluetooth and invites them
  to be DUs in the BMN.
• The construction of WTN User A initializes its
  soft AP and informs parts of the DUs via
  Bluetooth. The DUs selected connect to the soft
  AP via WiFi and become AUs.

17
Self-Organization Scheme

• The reason why only parts of DUs are selected to
  be AUs
• There exists an upper bound for the number of
  clients that can connect to the same soft AP at
  the same time.
• Severe transmission collisions would lead to the
  bottleneck effect of WiFi.

18
Cooperative Download Scheme

1. The initialization of download the URL of the
   target data is broadcasted to the AUs.
2. The process of download PU divides the data to
   be downloaded into small segments with constant
   size. The users download corresponding segments
   from the cellular network. For AUs the segment
   will be retransmitted to PU immediately.
3. The completion of download all the users upload
   their download records to the management server.
   Then their virtual credits are exchanged
   according to RAP scheme.

19
Cooperative Download Scheme

1. The initialization of download the URL of the
   target data is broadcasted to the AUs.
2. The process of download PU divides the data to
   be downloaded into small segments with constant
   size. The users download corresponding segments
   from the cellular network. For AUs the segment
   will be retransmitted to PU immediately.
3. The completion of download all the users upload
   their download records to the management server.
   Then their virtual credits are exchanged
   according to RAP scheme.

20
Cooperative Download Scheme

1. The initialization of download the URL of the
   target data is broadcasted to the AUs.
2. The process of download PU divides the data to
   be downloaded into small segments with constant
   size. The users download corresponding segments
   from the cellular network. For AUs the segment
   will be retransmitted to PU immediately.
3. The completion of download all the users upload
   their download records to the management server.
   Then their virtual credits are exchanged
   according to RAP scheme.

21
Cooperative Download Scheme

• Dynamic Task Dissemination Strategy
• The task to download the data is divided into
  small tasks, and each task requests a user to
  download a specified data segment. All the tasks
  are stored in a task queue.
• Once a PU/AU has finished downloading/retransmitti
  ng a segment, a new task will be assigned to it.
  This process keeps going on until the task queue
  is empty.
• Once a task is time-out, it will be added back to
  the head of the task queue.
• If an AU has no access to the Internet or it
  leaves the local network during downloading, the
  transmission blocked can still be recovered.
• New users are allowed to join the download
  process at any time.

22
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

22
23
Dynamic Self-Organization Strategy

• Power-saving Listening Method
• To reduce power consumption, we make the users in
  listening mode to be intermittent sleeping.
• E.g. with a period of 60 second, an user listens
  to the wireless channel for 30 seconds, and for
  the rest 30 seconds it keeps sleeping.
• Heuristic Scanning Method
• We control the interval between two adjacent
  scans of Bluetooth device discovery to control
  the power consumption in scanning mode.
• The scanning interval can be extended if the
  results do not change a lot in adjacent scanning
  rounds. Otherwise it will be reduced.

24
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

24
25
RAP Incentive Mechanism

• We propose the Reputation Adaptive Pricing Scheme
  (RAP) to
• Provide incentives for users who help others
  download data
• Prevent malicious users from downloading data
  only and not helping others relay traffics.
• The management server maintains a virtual credit
  account for each user as well as the download and
  relay history of the user.
• If a user wants to utilize our system, he must
  register an account and, in the meanwhile, he
  will get a certain amount of virtual credits as
  the initial money.

26
RAP Incentive Mechanism

• Stage 1 When a PU starts a co-download task and
  uploads the target data size, the server
  calculates the adapted total price for the
  download process
• where

punishment factor
size of the target data
flat unit price
total amount of data relayed in history
total amount of data downloaded in history
27
RAP Incentive Mechanism
 
positive constant


number of users in the download group
28
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

28
29
Measurement and Evaluation

• Measurement of Power Consumption in Listening
  Mode
• Measurement of Power Consumption in Scanning Mode

30
Measurement and Evaluation

• Evaluation of INDAPSON Download Rate
• The download rate increases linearly with the
  number of phones.
• With the maximum number of AUs controlled, the
  data rate of WiFi is always larger than that of
  3G, and the download data rate for PU is
  proportional to the sum of the 3G download rates
  of all the phones.

31
Measurement and Evaluation

• Simulation of RAP Incentive Scheme
• Our system mitigates the problem that the number
  of actions of download per day decreases
  dramatically at the end of a month, because users
  can leverage others surplus data plan and thus
  avoid being charged with a much higher price.

32
Measurement and Evaluation

• Simulation of RAP Incentive Scheme
• The different punishment factors have trivial
  influence on the number of transactions while the
  percentage of Malicious Users varies.
• with RAP is much less than that without
  RAP, and decreases as increases.

represents the average benefits that a
Malicious User gets every time he registers as a
new user.
33
Outline

• Introduction
• System Architecture
• Organization and Download Scheme
• Dynamic Self-Organization Strategy
• RAP Incentive Mechanism
• Measurement and Evaluation
• Summary

33
34
Summary

• We proposed a data plan sharing system based on
  self-organizing networks, which allows users to
  better manage their data traffic in a fixed data
  plan.
• In our system, with power-saving strategies
  applied, multiple wireless interfaces are used to
  construct a two-layer local network, which adapts
  to the dynamic characteristics of mobile
  terminals.
• Data are downloaded via multiple download links
  from the Internet, and the total download rate is
  improved.
• With RAP applied in our system, users are able to
  exchange their data traffic for credits by
  helping other users download data.
• Numerical experiments and simulations also show
  that our system can help users manage their data
  traffic effectively.

35
Thank you for listening