Improving Wireless Health Monitoring Using Incentive-Based Router Cooperation

1
Improving Wireless Health Monitoring Using
Incentive-Based Router Cooperation

• Upkar Varshney, Georgia State University
• ?????

2
Outline

• 1. Introduction
• 2. Incentive-based router cooperation
• 3. Analytical model
• 4. Performance evaluation
• 5. Conclusion

3
1. Introduction(1/5)

• In recent years, researchers have explored the
• requirements and potential benefits of
  wireless
• networks in health monitoring.
• One of the most difficult challenges associated
  with
• using wireless networks is the
  reliability of message
• delivery.
• End-to-end delays, frequency, and the number of
• patients also affect the quality of
  monitoring.

4
1. Introduction(2/5)

• Many healthcare providers are beginning to use
  infrastructure-oriented wireless networks such as
  IEEE 802.11 wireless LANs and cellular systems to
  monitor and track Patients.
• The infrastructure supports monitoring devices in
  hospitals and nursing homes, outdoors, and at
  home using base stations (cellular networks) or
  access points (wireless LANs).

5
1. Introduction(3/5)

• However, potentially spotty coverage due to time-
  and location-dependent channel quality and signal
  attenuation resulting in dead spots can
  significantly diminish message delivery
  reliability.

6
1. Introduction(4/5)

• Healthcare providers could use ad hoc networks to
  supplement the sometimes spotty and unreliable
  coverage of infrastructure-oriented wireless
  networks for patient monitoring.
• A proposed incentive-based approach encourages
  devices to cooperate as routers, significantly
  improving message delivery reliability.

7
1. Introduction(5/5)
8
2. Incentive-based router cooperation(1/9)

• Potential message delivery problems due to
  uncooperative routers can be overcome by
• Implementing persistent transmission, which lets
  uncooperative devices transmit at reduced power
  levels at reduced distances.
• Using more persistent multi-path routing schemes
  such as reliable multicast and reliable
  broadcast.
• Excluding uncooperative devices from the routes
  as in a multicast tree.

9
2. Incentive-based router cooperation(2/9)

• Perhaps the best way to obtain router cooperation
  is to offer a range of incentives from credits to
  higher-priority message delivery.
• Devices could decide whether to cooperate based
  on offered or earned incentives.
• Messages could be forwarded based on a devices
  history of cooperation.
• Devices that have been cooperative in the past
  would be given higher message-delivery priority,
  while those that have been uncooperative would be
  given lower priority.
• Higher incentives could also be assigned to
  certain types of messages, such as emergency
  alerts, to encourage their delivery.

10
2. Incentive-based router cooperation(3/9)

• Vital credits
• NTnetwork traffic load, depends on the number of
  monitored patients,
• frequency of monitoring, packets per
  monitored event, and routing
• scheme.
• Pmessage priority level
• Ccriticality of the routing device, as
  determined by its position in the
• network and the current routing scheme.
• L and M and Nconstants that can be chosen to
  emphasize certain factors
• in Vital credits
  that allow the modeling of different
• environments.

11
2. Incentive-based router cooperation(4/9)
12
2. Incentive-based router cooperation(5/9)
13
2. Incentive-based router cooperation(6/9)

• Cooperation Protocol
• Depending on the devices complexity and desired
  accuracy in decision making
• Continuous Value Cooperation Protocol (CVCP).
• Discrete Value Cooperation Protocol (DVCP).
• CVCP
• Larger ad hoc networks
• By keeping all parameters and computations at
  exact values
• Relatively accurate decision making
• More computation and storage overhead
• Suited to devices with complex functionalities

14
2. Incentive-based router cooperation(7/9)

• The health-monitoring devices running CVCP are
  likely to be more powerful and have functionality
  for power conservation.
• The devices can choose the values of constants A
  (stored credits) and B (available power) based on
  desired thresholds.

15
2. Incentive-based router cooperation(8/9)

• DVCP
• Smaller ad hoc networks
• Approximate values such as high (H), medium (M),
  and low (L)
• Simplifies computations
• Less overhead
• Less accurate decision making
• Simpler health-monitoring devices
• Not support sleep-cycle-based power conservation

16
2. Incentive-based router cooperation(9/9)

• For smaller incentive values at low network loads
  ,CVCP and DVCP performance should not differ
  greatly.
• CVCP is likely to achieve decision-making
  accuracy superior to DVCP in larger, more complex
  ad hoc networks.
• In general, for both CVCP and DVCP, the use of
  Vital credits should improve health-monitoring
  reliability.

17
3. Analytical model(1/8)

• Author developed an analytical model to assess
  the effectiveness of Vital-credits-based router
  cooperation in terms of message reliability under
  varying network load and size.
• To maintain tractability and reasonable accuracy,
  this model assumes
• The path between patient and healthcare
  professional is a pure multihop ad hoc network
• All devices have the same power budget and
  processing power
• The devices are uniformly distributed in the
  service area

18
3. Analytical model(2/8)
19
3. Analytical model(3/8)

• The probability of finding a path to a
  destination user is a joint probability of
  finding a minimum number of devices needed to
  form a path between a patient and a healthcare
  professional.
• Next-hop device probability
• UENext-hop device probability
• DEThe device distribution function (higher for
  uniformly low for cluster)
• NEThe number of cooperating devices
• CEThe device coverage
• ALength of service area
• BWidth of service area

20
3. Analytical model(4/8)

• UE-UN UE ( 1 PUN )
• UEUNNext-hop device probability with
  noncooperation
• PUNThe probability of encountering an
  uncooperative
• device
• ISTStored incentives
• IOFFOffered incentives
• L and MBe used to change the relative
  importance of
• stored and offered Vital
  credits
• FA factor used to normalize the probability to
  (0, 1) range

21
3. Analytical model(5/8)

• Message delivery reliability can be measured in
  terms of the probability of message reception.
• Unicasttransmitting messages to a single
• healthcare professional
• RU Reliability of message delivery using
  unicast
• routing
• H The number of hops to the destination

22
3. Analytical model(6/8)

• Anycasttransmitting messages to at least one
• healthcare professional
• RA Reliability of message delivery using
  anycast routing
• DN The number of healthcare professionals
  doing the
• monitoring

23
3. Analytical model(7/8)

• Multicastusing multiple paths to deliver
• messages
• RMReliability of message delivery using
  multicast
• routing
• MThe number of paths used

24
3. Analytical model(8/8)

• Broadcastusing every possible device to deliver
  messages
• RBReliability of message delivery using
  broadcast routing
• PBThe probability of finding a broadcast path to
  a definition
• PPThe number of paths between a patient and a
• professional

25
4. Performance evaluation(1/2)

• Assumed
• 3 healthcare professionals
• 18 devices
• The initial stored incentives were equal
• Offered incentives were based on network traffic
  and packet priority
• Device criticality was fixed at 1
• 9 for multicast routing
• 45 for broadcast routing.

26
4. Performance evaluation(2/2)
27
5. Conclusion(1/1)

• Ad hoc networks could be used to supplement the
  sometimes spotty and unreliable coverage of
  infrastructure-oriented wireless networks for
  health monitoring.
• The performance results from analytical model
  show that using incentives such as Vital credits
  can significantly improve message delivery
  reliability.