Integrating and Testing the Performance of the Wireless Sensor Networks Integrated into the C6 Virtual Reality Environment

1
Integrating and Testing the Performance of the
Wireless Sensor Networks Integrated into the C6
Virtual Reality Environment

• Bernard Lwakabamba

2
Outline

• Acknowledgments
• Introduction
• Current System
• Proposed Solution
• WSN Communication Architecture
• Mote Hardware
• Mote Software Environment
• Packet Delivery Issues Physical layer
• Packet Delivery Issues MAC sub-layer
• Packet Delivery Issues Error control sub-layer
• Conclusions
• References

3
Acknowledgments

• The CRCD project is sponsored by the NSF
  Engineering Education and Centers (EEC) project
  number 00-88071
• I would like to thank
• Dr. Dickerson for giving me the opportunity to be
  part of the CRCD project
• All of the students and professors that also
  participated in the CRCD project
• The faculty and staff of the Virtual Reality
  Application Center (VRAC)

4
Introduction

• The C6 is a fully enclosed VR environment,
  therefore all of the devices must be wireless.
• One of the goals for the NSF Combined Research
  Curriculum Development (CRCD) project was to
  design and implement a low-power wireless
  communication system for virtual environments.
• We utilized the emerging technology of Wireless
  Sensor Networks (WSNs), to provide a unified
  communication system between the sensing devices
  and the VR applications.

5
Current System
6
Proposed Solution
7
WSN Communication Architecture 46

• The current design utilizes a single-hop network
• Focus was placed on the physical and data link
  layers of the Motes.
• Physical layer
• Frequency selection, carrier frequency
  generation, signal detection, modulation, data
  encryption, etc.
• Data link layer MAC and Error Control sub-layers
• MAC sub-layer
• Channel allocation, collision avoidance,
  fairness, low-power listening
• Error Control sub-layer
• Detection and possible recovery of erroneous
  packets ( message reliability)

8
Mote Hardware Components
9
Motes Evolution2425

• The Mica2 and Mica2Dot Motes operating in the
  900 MHz ISM band were integrated into the C6 VR
  environment.

10
Mote Radio

• Radio ChipCon CC1000 RF transceiver
• FSK modulation
• Data rates 38.4 Kbps (def.) or 19.2 Kbps
  (Manchester encoding)
• Digitally programmable output power 5dBm to -20
  dBm
• Software programmable frequency 902 928 MHz
• 50 channels ( FCC reg.)
• 500 ft. LOS outdoor coverage range

Mica2Dot Mote
Mica2 Mote
11
Mote Microprocessor
Mica2Dot Mote
Mica2Dot Mote

• Microprocessor Atmega 128
• 4 MHz clock
• UART baud rate 19200 baud
• 6 10-bit ADC channels
• Vcc 3V
• Other
• 19 pin interface
• Battery 3V coin cell
• Single programmable LED

• Microprocessor Atmega 128
• 7.3728 MHz clock
• UART baud rate 57600 baud
• 8 10-bit ADC channels
• Vcc 3V
• Other
• 51 pin interface
• Battery 2 AA 1.5V batteries
• Three programmable LEDs

12
Mote Software Environment
13
TinyOS 2436

• Open-source event-driven OS developed by UC
  Berkeley, to support Mote WSN platforms
• TinyOS applications are written in nesC a C-like
  syntax programming language
• TinyOS uses a set of reusable system components,
  that link together to form an executable
• The component library contains sensor drivers,
  communication layer protocols, and DAQ tools
• Component types
• Tasks computational functions
• Event Handlers respond to hardware interrupts

14
UC Berkeleys Communications Stack (I)3637

• GenericComm AMStandard
• Defines Dest. Address, Handler ID, Group ID,
  Msg. length
• Hand off packet over the radio or UART Comm.
  Channel
• CC100Control
• Manages features of the radio
• Tx. power
• Tx. operating frequency
• Rx. Operating frequency
• Motes 902 928 MHz
• Channel spacing 500 kHz
• -20 dBm Tx. power 5 dBm

TinyOS v.1.1.0 release
15
UC Berkeleys Communications Stack (II)40

• CC1000RadioIntM
• Byte level interface between MCU and radio
• Provides Tx. and Rx. Data transfer using the
  default CSMA/CA scheme
• Tx. data
• Carrier sensing using RSSI and preamble from
  other nodes (SPI port)
• Random back off mode if nodes senses a signal
• Rx. data
• Preamble detection, synchronization, CRC check (
  SPI port)
• Reject packet bad CRC, or mismatched Group ID
• Low-power listening
• Timer components toggle Tx. and Rx. periods

16
Default TinyOS Message Structure38

• Default packet length 36 bytes ( 288 bits)
• Header
• Dest. address 2 bytes ( Radio/UART)
• Handler ID 1 byte
• Group ID 1 byte
• Msg. length 1 byte
• Payload
• Source address 2 bytes ( Node ID)
• Sample counter 2 bytes
• ADC channel 2 bytes
• ADC data readings 10 readings _at_ 2 bytes each

17
BMAC ( Evolved MAC protocol)41

• Default contention based MAC for TinyOS v1.1.3
  and beyond
• Provides several features as parameters to the
  upper layers of the network stack ( developers
  trade-off)
• Preamble length
• Left at 5 bytes, TinyOS v1.0 28 bytes
• Modified back-off mechanism
• Modified initial and congestion back-off
  mechanisms

18
ECC Communications Stack 3031

• Mirrors the network stack of the older Mica
  Motes, modified for the Mica2Dot Motes
• RadioCRCPacket
• Tx. Calculates CRC of Tx. data and appends it as
  a trailer packet
• Rx. Calculates CRC and checks for mismatches
• RFCommM
• Controls settings and parameters of radio
• ChannelMonEccC/ ChannelMonC
• Carrier sensing, preamble packet detection
• ChannelMonEccC calls SecDedEncoding the ECC
  module

TinyOS v.1.0 modification
19
Packet Delivery Issues Physical Layer
20
Experiments Setup

• Experimental Model
• One event and one base station node connect to
  the CPU.
• Maximum distance 25 feet
• Opt. freq. 918 MHz
• Performance metrics
• Packet acceptance rate (PAR)
• RSSI
• Communication range
• Experiments
• In C6 with a LOS, vary Tx. power and distance.
• In C6 w/o a LOS, propagate through users body (
  fixed Tx. power and distance).
• Look at relationship between RSSI and PAR.

21
LOS Varying distance and Tx. power

• Parameters
• Event sensor transmit at a rate of 1 pkts/sec
• PAR calculated by disregarding the corrupted
  packets that do not pass the CRC and Group ID
  checks. 46
• Results obtained after 250 packets were received.
• Conducted three times and took the average.
• Comments
• VR applications require a highly reliable comm.
  system
• High PERs increase latency and degrades user
  interaction
• Trade-off between energy efficiency and
  performance
• Developer must determine acceptable PERs
• Multi-hop network use lower Tx. power and more
  relay sensor nodes

22
Non-LOS Experiments

• Current design will require the base station
  Mote to be attached to the wearable CPU on a
  user worn back pack.
• Event sensor nodes will have to propagate
  through the users body.
• Parameters are the same as the LOS experiments,
  except will only consider Tx. power of 5 dBm, and
  distance of 5 feet.

23
Received Signal Strength Experiments (I)

• Look at the cross layer interactions between the
  physical layer and the upper layer of the Mote
  network stack
• The MAC sub-layer of the Motes uses a carrier
  sensing mechanism to determine channel state
• The CC1000 transceiver generates an analog signal
  (ADC channel 0), to determine the strength of the
  received signal
• The signal ranges from 0 to 1.2 V ( 0 strongest
  possible signal strength)
• In the 900 MHz ISM band the RSSI is calculated by
  34

• The theoretical noise floor/ receiver
  sensitivity is 105.5 dBm
• Need to determine if there is a direct
  correlation between the RSSI and PAR.

24
Received Signal Strength Experiments (II)

• Setup
• In the Anechoic Chamber
• 5dBm Tx. power ( new battery)
• Data rate 15 pkts/sec
• Opt. freq. 918 MHz
• Dist. 5 feet
• Used the HP 8594E Spectrum Analyzer (9 kHz to
  2.9 GHz), to determine the RSS.
• Nodes rotated in 10 degree intervals until all 36
  readings were obtained.
• Two experiments
• LOS
• Non-LOS Human propagation
• Note
• 1 foot -38.25 dBm

25
Received Signal Strength Experiments (III)

• Comments
• High RSSI may not guarantee a high PAR, but a low
  RSSI will usually relate to a low PAR for the
  sensor nodes
• Event sensor node may be transmitting data but
  the base station node may detect it as noise/
  corrupted data ( RSSI ltlt noise floor)

26
Packet Delivery Issues MAC Sub-layer
27
MAC features and Tradeoffs for the C6

• Collision Avoidance
• Collision are energy inefficient and decrease
  data throughput
• Retransmission degrade user interaction in VR
  applications
• Latency
• Def. Total delay time between user-enforced
  action and system response displayed on VR
  application.
• Latency degrades user interaction, and can cause
  cyber sickness
• Protocol Overhead
• Control packets and MAC headers required to
  synchronize reception and transmission of payload
  data
• Long protocol overhead viewed as energy
  inefficient and they decrease the channel
  utilization
• Low Power Listening
• Idle listening and overhearing energy
  inefficient
• Low power listening enforce a low duty cycle to
  increase network lifetime ( sleep, idle, active
  modes)
• Tradeoff will increase latency and decrease
  max. achievable data rate

28
Fairness and Throughput (I)

• Setup
• Each sensor node was assigned a unique node ID.
• 5 foot radius
• Three experiments were conducted at various
  workloads
• 4 pkts/sec
• 10 pkts/sec
• 15 pkts/sec
• Each experiment was conducted three time to
  obtain an average of the results

29
Fairness and Throughput (II)
30
Fairness and Throughput (III)
31
Fairness and Throughput Comments

• Results are crucial for VR application developers
  to understand especially when several sensor
  nodes will be utilized
• Configuring nodes to transmit at higher packet
  rates can reduce the overall system latency, but
  it degrades the ability to utilize multiple nodes
  in the same channel
• Ideas
• Need to interface more sensor onto the same event
  sensor node
• Use more than one channel for each user
• Alternative
• Developers determine minimum packet rate that
  would still allow user to effectively interact
  with the VR app.
• Will not have to use higher packet rates which
  may saturate channel

32
Conclusions

• Physical layer
• It was determined that the 5 dBm was optimal
  transmit power
• Reduces network lifetime but improves PAR and
  reliability
• To use lower output power might require the use
  of a multiple hop network
• RSSI drops when signal has to propagate through
  the user, could consider changing the location of
  the base station node or multi-hop network
• MAC sub-layer
• Multiple users can operate in the same channel,
  different Group IDs, need to determine maximum
  threshold
• Reducing overhead control packets and back off
  mechanism can increase the throughput and channel
  utilization
• Workload affected BMACs channel allocation and
  fairness capabilities
• More sensors on node

33
Conclusions

• MAC sub-layer (cont.)
• More than one channel per user
• Developers need to be smart when the program
  the nodes
• Could look at other emerging MAC protocols for
  WSNs, SMAC 43, PRIME 42.
• Error control sub-layer
• Need to find a low complexity coding scheme that
  can correct multiple or burst bit errors
• Accurate models of the Motes network stack can
  help to determine performance of coding schemes
• Hybrid schemes might be useful but they may
  require over the air reprogramming

34
References