FBRT: A Feedback-Based Reliable Transport Protocol for Wireless Sensor Networks

1
FBRT A Feedback-Based Reliable Transport
Protocol for Wireless Sensor Networks
1st Year MPhil Presentation

• Yangfan Zhou
• November, 2004
• Supervisors Dr. Michael Lyu and Dr. Jiangchuan
  Liu

2
Presentation Outlines

• 1. Introduction
• 2. Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

3
Presentation Outlines

• 1. Introduction
• 2. Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

4
Introduction

• Wireless Sensor Networks (WSN)
• Sensors nodes measure physical phenomena.
• Target tracking
• Environment data measurement
• Engineering measurement
• Sensor nodes form an ad-hoc multi-hop wireless
  network to convey data to a sink.

5
Introduction

• WSN Challenges
• WSN suffers from energy constraint
• WSN condition
• Unreliable wireless link
• High packet loss rate
• Network Dynamics
• Node failures
• Link failures
• Dynamic traffic load

6
Introduction

• Reliable sensor-to-sink data transport for WSN
• It is Important
• Objective
• to assure that the sink can receive desired
  information is very important.
• The work presented here is to address this
  problem.

7
Introduction

• Reliable sensor-to-sink data transport for WSN
• 100 reliable data transport is not necessary.
• Reliability means desired information has been
  achieved
• Source sensors might have different contributions
  

8
Introduction

• Reliable sensor-to-sink data transport for WSN

Bias the transport scheme
9
Introduction

• Current Approaches on WSN data transport
• RMST Reliable Multi-Segment Transport by
  Heidemann et al, SNPA03
• PSFQ Pump Slowly, Fetch Quicklyby C. Wan et al,
  WSNA02
• Not applicable for sensor-to-sink data transport

10
Introduction

• ESRT Event to Sink Reliable Transport by
  Sankarasubramaniam et al, MobiHoc03
• Congestion detection
• Queue Length
• Reliability consideration
• Receiving rate of the incoming packets
• Rate adjustment
• Unbiased adjustment

11
Introduction

• CODA Congestion Detection and Avoidance by C.
  Wan, SenSys'03,   
• Congestion detection
• channel sampling
• Congestion avoidance
• Slowing down the sending rate
• It has not addressed the reliability issues.

12
Presentation Outlines

• 1. Introduction
• 2. Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

13
Motivations

• Issues to be addressed to provide reliable
  sensor-to-sink data transport
• Source reporting rate adjustment scheme
• Routing scheme

14
Design Considerations

• Reporting Rate Control
• Relationship between receiving rates and
  distortion
• Different contributions of source nodes.
• Different energy costs for communication.
• Rate control scheme should employ an optimization
  approach to minimize energy consumption of the
  WSN.
• Adjust the rates so that energy consumption is
  minimized subjected to that the distortion is in
  a given range.

15
Design Considerations

• Distortion and Sensor Contribution
• Application Specific, should be determined by
  applications.
• Rate Control
• Cooperation of the application and the transport
  protocol.

Figure
16
Design Considerations

• Communication cost estimation
• Hop number from the source to the sink
• Simple
• Inaccurate
• Node Price
• Our metrics Total number of packets sent by the
  in-network nodes for per packet received by the
  sink
• Accurate
• Physical layer overhead
• But hard to implement

17
Design Considerations

• Node Price

NP(x) Node price of X node
ns downstream neighbors Perc(i) the percentage
of traffic that is routed to node i
The hop loss rate between node n and
node i The loss rate of
the path from node i to the sink
18
NP(sink) 0 PathLossRate(Sink) 0
Sink
PathLossRate(2)
PathLossRate(3)
2
NP(2)
NP(3)
3
HopLossRate(2)
HopLossRate(3)
Perc(2)
Perc(3)
1
19
Design Considerations

• Node Price Estimation
• Each node can calculate its NP and PathLossRate
  based on
• The feedback of NP and PathLossRate of its
  downstream neighbors
• The HopLossRate to each of its downstream
  neighbors
• The routing scheme Perc(i)
• Two unknowns
• The HopLossRate
• The routing scheme (Discussed Later)

20
Design Considerations

• Hop Loss Rate
• mainly caused by three factors
• Congestion
• Signal Interference
• Fading.
• packet loss rate will exhibit graceful increasing
  behavior as the communication load increases
  (IEEE 802.11 MAC)
• reasonable to estimate the packet loss rate based
  on an exponential weighted moving average (EWMA)
  estimation approach.

21
Design Considerations

• Accurate and Current Hop Loss Rate Estimation
• Indicates the congestion condition well
• Indicates the weak link well
• Node Price based on loss rate estimation
• Indicates the dynamic wireless communication
  condition from the node to the sink well
• can help to determine the reporting rates
• can help to determine the routing scheme

22
Design Considerations

• Routing Schemes
• Minimizing local NP.
• Locally optimal energy consumption, minimizing
  the energy consumed for the sink to receive per
  packet from me)

2
NP(2)
NP(3)
3
HopLossRate(2)
HopLossRate(3)
Perc(2)
Perc(3)
1
23
Design Considerations

• Routing Schemes Oscillation Avoidance

24
Analysis

• Routing Schemes Oscillation Avoidance
• Gradually shift traffic to best path
• Adaptive to downstream dynamics

2
NP(2)
NP(3)
3
HopLossRate(2)
HopLossRate(3)
Perc(2)
Perc(3)
1
25
Presentation Outlines

• 1. Introduction
• 2. Motivations and Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

26
Protocol Implementation

• Task assignment Broadcast interest packet
• Get possible downstream neighbor information
• Select path with the lowest hop number to the
  sink as tentative best path
• Low reporting rate requirement tentatively

27
Protocol Implementation

• Link loss rate estimation
• Measured according to packet serial numbers holes
• Estimated with an EWMA approach.

28
Protocol Implementation

• Feedback of communication condition
• Checking the following parameters in a given
  interval
• A node NP
• A nodes path loss rate to the sink
• Link loss rate from upstream neighbors
• If they are changed, feed back the new value to
  upstream nodes
• higher priority.

29
Protocol Implementation

• Feedback of newly desired reporting rates

Application
Application
Rate adjustment
Sensor Data
Rate adjustment feedback
Sensor Data Source NP
FBRT
FBRT
FBRT
Node
Encapsulate my NP into data packets
The Sink
Source
30
Presentation Outlines

• 1. Introduction
• 2. Motivations and Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

31
Simulation results

• Coding FBRT over NS-2
• Setting of the network
• Scheme 1 Based on directed diffusion with ESRT
  scheme. ()
• Scheme 2 FBRT (o)

Area of sensor field 1500m1500m
Number of sensor nodes 100
MAC IEEE 802.11 without CTS/RTS and ACK
Radio power 0.2818
Packet length 36 bytes
Transmit Power 0.660 W
Receive Power 0.395 W
Feedback interval 1 second
IFQ length 50 packets
Simulation Time 1000 seconds
32
Simulation results

• Simulation Network

33
Simulation results

• Results

Energy consumed of the WSN (J)
34
Simulation results

• Results

35
Presentation Outlines

• 1. Introduction
• 2. Motivations and Design Considerations
• 3. Protocol Implementation
• 4. Simulation Results
• 5. Conclusion

36
Conclusion

• we propose FBRP, a feedback-based protocol to
  address reliable sensor-to-sink data transport
  issue
• FBRP optimizes the energy consumptions with two
  schemes.
• the sink's rate control scheme that feeds back
  the optimal reporting rate of each source.
• the locally optimal routing scheme for in-network
  nodes according to the feedback of downstream
  communication conditions.
• Simulation results verify its effectiveness for
  reducing energy consumption.

37

• Thank You