mạng cảm biến không dây

HANOI UNIVERSITY OF SCIENCE AND TECHNOLOGY SCHOOL OF INFORMATION & COMMUNICATION TECHONOLOGY ─────── *  ─────── THESIS SUBMITTED FOR PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR DEGREE OF ENGINEER IN INFORMATION TECHNOLOGY REAL-TIME TRACKING WIRELESS SENSOR NETWORK WITH ENERGY EFFICIENT MAC PROTOCOL Author: NGUYEN TRUNG QUAN Class ICT-541 Supervisor: Dr NGO QUYNH THU HANOI 5-2014 REQUIREMENTS  FOR  THE  THESIS Student information Student name: NGUYEN TRUNG QUAN Tel: 01666980501 Email: quannt24@gmail.com Class: ICT-541 Program: ICT This thesis is performed at: Hanoi From: 24/2/2014 To: 30/5/2014 Goal of the thesis  Design  a  tracking  system  basing  on  wireless  sensor  network  with  high  accuracy,  low   delay  and  energy  efficient  Design   a   MAC   protocol   for   the   system   which   can   support   low   delay   and   energy   efficiency  requirements Main tasks  Study  the  foundation  of  wireless  sensor  networks  and  its  main  characteristics  Study  some  related  tracking  system  and  protocols  in  wireless  sensor  networks  Study   about   tracking   algorithms   and   their   applicability   in   tracking   wireless   sensor   network  Propose   a   design   for   the   system   and   implement   its   simulation   in   OMNeT++   environment  Propose  a  MAC  protocol  which  can  achieve  low  delay  and  energy  efficiency  while   maintaining  decent  packet  delivery  rate  Carry  out  experiments  on  simulation  and  analyze  results Declaration of student: I – Nguyen Trung Quan – hereby warrant that the Work and Presentation in this thesis are performed by myself under the supervision of Dr Ngo Quynh Thu All results presented in this thesis are truthful and are not copied from any other work Hanoi, 30 May 2014 Author Nguyen Trung Quan Attestation of the supervisor on the fulfillment of the requirements of the thesis: Hanoi, 30 May 2014 Supervisor Dr Ngo Quynh Thu This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 ACKNOWLEDGEMENT First   of   all,   I   would   like   to   thank   Dr   Ngo   Quynh   Thu   (Department   of   Data   Communications   and   Computer   Networks   –   School   of   Information   and   Communication   Technology)   and   Dr   Tran   Quang   Vinh   (School   of   Electronics   and   Telecommunications),   who  have  supervised  and  oriented  me  throughout  this  research  I  also  would  like  to  thank   lecturers  of  Hanoi  University  of  Science  and  Technology  who  have  educated  and  provided   me  knowledge  for  carrying  out  this  research  I  would  also  like  to  thank  the  fellows  in  my   research  lab,  who  have  supported  me  by  their  valuable  advices  and  inspired  me  with  their   creative  ideas Finally,  with  all  of  my  gratitude,  I  would  like  to  send  my  thanks  to  my  family  Without   their  continuous  support  and  encouragement,  I  will  never  complete  this  research This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 iii FOREWORD Nowadays,   wireless   sensor  networks   have   been  developed   extensively   in  many   fields   of   life,   many   application   of   them   have   been   deployed   in   reality,   such   as   monitoring   environment  and  weather,  military,  objects  tracking,  traffic  monitoring  and  controlling,  etc   The   major   aim   of   this   research   is   designing   a   wireless   sensor   network   for   tracking   application,  which  can  provide  accurate  results  in  real-time  and  have  high  energy  efficiency   The  design  should  have  a  robust  and  versatile  tracking  process  with  appropriate  algorithms   In  addition,  an  efficient  communication  scheme  is  crucial,  not  only  for  low  delay  and  high   energy   efficiency   but   also   tracking   accuracy   In   this   research,   a   new   design   for   MAC   protocol   is   proposed,   which   can   cooperate   with   the   routing   protocol   EMRP   to   provide   a   low-delay  and  energy  efficient  communication  basis A   simulation   of   the   system   is   implemented   in   OMNeT++   simulation   environment   to   analyze  the  working  of  the  system  and  evaluate  efficiency  of  the  design This   research   is   guided   and   supervised   by   Dr   Ngo   Quynh   Thu   and   Dr   Tran   Quang   Vinh   To   develop   the   design,   I   also   referred   to   many   related   research   results   of   other   authors,  some  of  which  I  have  modified  and  adapted  to  the  design The  content  of  this  thesis  is  structured  as  follow:  Part  1:  Problem  statement  and  orientation  for  solution o Chapter   I:   This   chapter   provides   some   fundamentals   about   wireless   sensor   networks  and  their  specific  characteristics  An  overview  about  main  components   of  a  tracking  wireless  sensor  network  is  also  introduced o Chapter  II:  This   chapter  contains  a  more  specific  statement   about  the  problem   should   be   solved   about   developing   an   accurate   and   efficient   tracking   wireless   sensor  network  along  with  an  orientation  for  solution   In  addition,  a  new  lowdelay  and  energy  efficient  MAC  protocol  is  proposed  as  a  part  of  the  system  Part  2:  Research  results o Chapter   III:   This   chapter   provides   a   detailed   design   for   the   required   system   Structure  of  the  system,  cooperation  of  components  and  detailed  description  of   the  proposed  MAC  protocol  are  stated  in  this  chapter o Chapter   IV:   In   this   chapter,   a   simulation   of   the   proposed   system   will   be   introduced  Simulation  results  will  be  recorded  and  analyzed This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 iv TÓM  TẮT  NỘI  DUNG  ĐỒ  ÁN  TỐT  NGHIỆP Mạng  cảm   biến  không  dây  (WSN)  là  một   hướng  nghiên  cứu  mới,  mở  ra  cơ  hội   phát   triển  các  ứng  dụng  trong  nhiều  lĩnh  vực  của  đời  sống  WSN  được  cấu  thành  từ  những  nút   mạng  là  các  thiết  bị  có  kích  thước  nhỏ  nhưng  có  năng  lực  tính  toán  độc  lập,  khả  năng  truyền   thông  và  giám  sát  môi  trường  xung  quanh  Các  nút  mạng  được  kết  nối  với  nhau  qua  các  kết   nối  không  dây  và  hoạt  động  một  cách  độc  lập  trong  môi  trường  mở  và  thường  có  ít  hoặc   không  có  sự  bảo  trì  của  con  người  Chúng  được  gọi  là  các  nút  cảm  biến  và  thường  có  giá   thành   rẻ   Nhờ   vậy,   WSN   đem   lại     phương   cách     cho     nhiệm   vụ   giám   sát   môi   trường  dựa  vào  các  dữ  liệu  cảm  biến  Dựa  trên  WSN,  con  người  có  thể  phát  triển  nhiều  ứng   dụng  trong  các  lĩnh  vực  quân  sự,  chăm  sóc  sức  khỏe,  giám  sát  thảm  họa… Nội  dung  của  đồ  án  chỉ  tập  trung  nghiên  cứu  phát  triển  một  hệ  thống  mạng  cảm  biến   không  dây  phục  vụ  cho  ứng  dụng  theo  dõi  mục  tiêu  Chức  năng  chính  của  hệ  thống  là  theo   dõi  sự  di  chuyển  của  mục  tiêu  trong  khu  vực  được  triển  khai  mạng  Cùng  với  đó,  hệ  thống   cần  phải  có  độ  chính  xác  cao  trong  kết  quả  theo  dõi,  có  khả  năng  làm  việc  trong  thời  gian   thực  và  duy  trì  hiệu  quả  cao  về  sử  dụng  năng  lượng  Cùng  với  việc  thiết  kế  hệ  thống  và  cài   đặt  trên  OMNeT++,  giao  thức  đa  truy  cập  XT-MAC  được  đề  xuất,  có  nhiệm  vụ  đóng  vai  trò    trong  việc  tăng  cường  hiệu  quả  sử  dụng  năng  lượng  trong  khi  vẫn  duy  trì  độ  trễ  thấp    tỉ  lệ  gửi  gói  tin  cao  Các  mô  phỏng  chi  tiết  đã  được  thực  hiện  nhằm  đánh  giá  thiết  kế  của   hệ  thống  về  các  tiêu  chí  độ  chính  xác,  hiệu  quả  sử  dụng  năng  lượng,  độ  trễ  và  tỉ  lệ  gửi  gói   tin  thành  công This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 v ABSTRACT Wireless   sensor   network   (WSN)   is   a   new   approach   for   research   which   opens   opportunity  for  development  of  many  applications  in  vast  of  fields  A  WSN  is  a  network  in   which   each   node   is   small   in   size   but   has   independent   computing,   communicating   and   environment   monitoring   capabilities   They   are   connected   by   wireless   channels   and   distributed   in   open   environment   where   they   operate   autonomously   with   few   or   even   no   maintenance   of   human   These   nodes   are   called   sensor   nodes   and   usually   come   with   low   price  Because  of  that,  WSN  provides  a  versatile  means  for  monitoring  environment  basing   on  sensed  data  from  sensor  nodes  Basing  on  WSN,  people  can  develop  many  applications   in  military,  health  care,  disaster  monitoring,  etc This   thesis   only   concentrates   the   research   in   developing   a   WSN   system   for   object   tracking   application   Main   function   of   the   system   is   tracking   movement   of   objects   in   network   area   In   addition,   the   system   should   produce   good   accuracy   in   real-time   when   maintaining  decent  energy  efficiency  Along  with  the  system  design  and  implementation  in   OMNeT++,  XT-MAC  –  a  multiple  access  channel  protocol  is  proposed,  which  undertakes   the  main  role  in  improving  energy  efficiency  while  maintaining  low  delay  and  high  packet   delivery  rate  Extensive  simulations  are  also  implemented  to  evaluate  the  design  of  tracking   system  in  terms  of  tracking  accuracy,  energy  efficiency,  latency  and  packet  delivery  ratio This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 vi CONTENTS REQUIREMENTS FOR THE THESIS ii ACKNOWLEDGEMENT iii FOREWORD iv TÓM TẮT NỘI  DUNG  ĐỒ ÁN TỐT NGHIỆP v ABSTRACT vi LIST OF TABLES ix LIST OF FIGURES x ACRONYMS xi PART 1: PROBLEM STATEMENT AND SOLUTION ORIENTATION .1 CHAPTER I: THEORETICAL BASIS I.1 Introduction of wireless sensor networks .1 I.1.1 General information .1 I.1.2 Characteristics of wireless sensor networks I.1.3 Problems in wireless sensor networks I.1.4 Applications of wireless sensor networks I.2 Target tracking in wireless sensor network I.2.1 Related work – Energy efficient MAC protocols I.2.2 Related work – Routing in WSNs I.2.3 Related work – Positioning methods .7 I.2.4 Developing an accurate and efficient tracking WSN 10 CHAPTER II: PROBLEM STATEMENT AND ORIENTATION FOR SOLUTION 11 II.1 Problem introduction 11 II.2 Problem statement .12 II.2.1 Expected features 12 II.2.2 Inputs 12 II.2.3 Outputs .12 II.2.4 Assumptions .12 II.3 Orientation for solution .12 PART 2: RESEARCH RESULTS 14 CHAPTER III: SYSTEM DESIGN 14 III.1 System models 14 This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 vii III.1.1 Network model 14 III.1.2 Model of physical layer .14 III.1.3 Sensing model 16 III.2 Tracking system design 16 III.2.1 Protocol stack 17 III.2.2 Design of XT-MAC protocol 21 III.2.3 Tracking process .24 CHAPTER IV: SIMULATIONS AND ANALYSES 30 IV.1 Simulation environment 30 IV.1.1 Introduction to OMNeT++ simulation framework 30 IV.1.2 Modeling concepts 31 IV.1.3 System simulation .33 IV.2 Experiments .34 IV.3 Result analyses 36 IV.3.1 Tracking accuracy .36 IV.3.2 End-to-end delay .41 IV.3.3 Energy consumption 42 IV.3.4 Packet delivery rate of data link layer 43 CONCLUSION AND FUTURE DEVELOPMENT 45 REFERENCES 46 This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 viii LIST  OF  TABLES Table IV-1: Parameters in simulation .36 Table IV-2: Tracking errors statistics .41 Table IV-3: End-to-end delay statistics 42 Table IV-4: Packet loss caused by data link layer 43 Table IV-5: Delivery rate of relayed packets 44 This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 ix LIST  OF  FIGURES Figure I-1: Sample wireless sensor network .1 Figure I-2: Components of tracking system Figure I-3: Ideal measurements for Lateration method Figure III-1: Radio state machine of CC2420 transceiver 15 Figure III-2: Simplified radio state machine .16 Figure III-3: Tracking process 17 Figure III-4: Protocol stack .18 Figure III-5: Strobe sending timeline 22 Figure III-6: Sending workflow of XT-MAC 23 Figure III-7: Sensing cycle adjustment (node broadcasts SYNC_REQUEST) .25 Figure III-8: Sensing cycle 27 Figure III-9: Trace collection process .29 Figure IV-1: Simple and compound modules 31 Figure IV-2: Raw positioning data delivered with XT-MAC 37 Figure IV-3: Raw positioning data delivered with B-MAC .37 Figure IV-4: Filtered traces with XT-MAC 38 Figure IV-5: Filtered traces with B-MAC .38 Figure IV-6: Tracked path (without timestamps) .39 Figure IV-7: Positioning error produced at CH 40 Figure IV-8: Tracking error produced at BS (including jump points) 40 Figure IV-9: End-to-end delay 42 Figure IV-10: Total residual energy 43 Figure IV-11: Residual energy maps at end of simulation .43 This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 x Parameters   can   be   used   to   customize   simple   module   behavior,   and   to   parameterize  the  model  topology Parameters  can  take  string,  numeric  or  boolean  values,  or  can  contain  XML  data   trees   Numeric   values   include   expressions   using   other   parameters   and   calling   C   functions,   random   variables   from   different   distributions,   and   values   input   interactively  by  the  user Numeric-valued   parameters   can   be   used   to   construct   topologies   in   a   flexible   way  Within  a  compound  module,  parameters  can  define  the  number  of  submodules,   number  of  gates,  and  the  way  the  internal  connections  are  made IV.1.3 System simulation Basing  on  the  environment  provided  by  OMNeT++,  the  simulation  of  researched   system  is  constructed  with  the  following  components: a Base framework This   is   the   backbone   of   the   simulation   system,   providing   basic   structures   and   interfaces   for   more   detailed   components   (derived   modules)   This   component   contains  following  sub-components:  ChannelMgr:   This   is   the   central   module   managing   channel   state   of   whole   network  (at  every  node),  organize  it  into  a  directed  graph  data  structure  where   each   node   (graph   node,   not   network   node)   is   an   object   storing   channel   information   corresponding   to   an   network   node;;   and   each   edge   represents   a   physical  connection  between  two  nodes  For  example  A  →  B  represents  “B  is   in  transmission  range  of  A”  Entities:  Interface  for  all  entities  in  the  simulation,  such  as  sensor  nodes,  base   station  and  moving  targets  Generic  messages:  Generic  formats  for  messages  used  in  simulation,  such  as   packets  used  in  each  layer  or  command  messages  for  module  interaction  in  a   node  Base   modules:   Provide   interface   and   basic   logic   for   full   featured   modules   representing  components  of  entities  in  simulation  All  specific  modules  must   derive  from  these  modules  Base  networks:  Basic  interfaces  for  constructing  a  network b Full featured modules These  modules  have  fully  implemented  logic  and  are  submodules  of  entities:  Application   layer:   application   layer   (of   BS   or   sensor   nodes)   implementing   tracking  functions  Network   layer:   implementation   of   EMRP   routing   protocol   More   specific   packet  formats  are  also  defined This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 33  Data  link  layer:  implementations  of  XT-MAC  and  B-MAC  Physical   layer:   simulate   physical   layer   compliant   with   IEEE   802.15.4   standard  Energy:  simulate  power  supply  of  network  nodes  Mobility:  manage  position  of  entities  in  simulation  area  Other  modules:  sensor  device  and  signal  generator  simulations c Entities This   component   contains   simulated   specific   entities   in   network   such   as   sensor   nodes,   base   station   and   moving   targets   Each   entity   contains   full   featured   submodules  and  therefore  can  be  used  in  simulation  for  researching d Networks This  component  contains  description  for  simulated  network  setups e Helpers This  component  contains  modules  which  are  not  present  in  real  world;;  however,   these  modules  support  simulation  process  and  helps  simplifying  the  implementation   of   the   simulation   They   provide   some   functions   such   as   statistic   collection,   node   arrangement  or  graphic  decoration,  etc f Utilities This   component   contain   common   used   libraries   and   objects   which   is   not   provided  by  OMNeT++,  such  as  library  for  calculating  matrices,  Gaussian  noise,  etc IV.2 Experiments The   simulation   is   developed   in   order   to   evaluate   tracking   quality   produced   by   the   whole   system   Besides   that,   to   inspect   performance   of   the   proposed   MAC   protocol   more   clearly,   B-MAC   is   also   implemented   as   reference   There   are   two   configurations  for  experiments  are  performed,  one  with  XT-MAC  as  MAC  protocol   of   network   nodes   (including   BS)   and   the   other   with   B-MAC   Except   specific   configuration   parameters   of   the   two   MAC   protocols,   all   other   configuration   parameters   of   all   layers   in   protocol   stack   are   kept   unchanged   The   detail   configuration   is   showed   in   Table   IV-1   To   evaluate   performance   of   two   simulated   systems,   data   about   tracking   accuracy,   end-to-end   delay,   energy   consumption   and   packet  delivery  rate  is  collected  and  examined  carefully In  simulation,  the  simulated  network  is  configured  in  compliant  with  the  stated   network  model  More  concretely,  there  are  256  sensor  nodes  evenly  distributed  over   an  area  of  400*400  m2  A  base  station  is  positioned  at  the  coordinates  of  (200,  400)   (unit  is  meter)  One  target  moves  in  the  network  area  after  ten  second  from  start  of   simulation;;  the  target’s  speed  ranges  from  6  to  12  m/s  Sensing  range  of  the  target  is   35   m;;   in   reality   this   sensing   range   depends   on   original   intensity   of   signal   radiated   This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 34 from  target  and  sensitivity  of  sensors,  however,  with  stated  assumptions,  the  sensing   range  is  represented  in  distance  in  the  simulation  for  simplicity  Measurement  error   has  standard  distribution  with  expected  value  is  0  m  and  standard  deviation  is  15%   of   sensing   range   Each   network   node   has   transmission   range   of   40   m   (again,   this   representation   is   for   simplicity)   Sensor   nodes   work   with   length   of   each   sensing   cycle   is   0.5   s  At   start   of   simulation,   every   sensor   node   is   provided   with   identical   amount   of   energy;;   because   length   of   tracking   scenario   is   short   (about   several   minutes),  the  initial  energy  of  each  node  will  be  set  to  a  small  value  of  5  mWh  so   that  we  can  examine  the  energy  efficiency  of  the  system   more  clearly  Base  station   will  have  infinite  power  supply Parameter Simulated area Number of nodes Sensing range Standard deviation of measurement error Initial energy capacity Bit rate Transmission range Transceiver’s  delay for switching to RX or TX mode Transceiver’s  delay for switching to IDLE mode Power consumption in IDLE mode Power consumption in RX mode Power consumption in TX mode Unslotted CSMA/CA parameters EMRP’s  initialization  length EMRP’s  switching  energy Sense period (length of sensing cycle) Time for CH broadcasts beacon message for sensing synchronization CDF’s   Distance threshold for collecting target position for a trace Distance threshold for collecting target position for a trace XT-MAC strobeTime XT-MAC reservedInterval XT-MAC strobePeriod XT-MAC listenInterval XT-MAC sleepInterval XT-MAC timeout to stay in ACTIVE state Value 400 * 400 m2 256 35 m 0.15 * 35 m mWh 250 kbps 40 m 12 symbol periods = 0.000192 s 0s 1.278 mW 56.4 mW 52.2 mW Default values specified for 802.15.4 standard First 10 seconds of simulation 0.5 mWh 0.5 s At 0.45 s in each cycle of CH node 0.7 30 m 2s 0.003744 s 0,003744 s 0.008768 s 0.011232 s 0.15 s 1s This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 35 B-MAC checkInterval 0.01 s B-MAC preamble length 323 bytes B-MAC timeout to wait for packet 1s after sense activity in channel Table IV-1: Parameters in simulation The   simulation   finishes   when   the   target   completes   its   course   Outputs   and   statistical  data  include:  Tracking  output  can  be  outputted  in  real-time  but  may  contain  junk  traces  and   is  not  stored  as  final  output  of  the  system  Post-processed  output  can  be  produced  with  a  constant  lag  of  junk  traces’  old   threshold   compared   to   previous   real-time   output;;   however   this   output   is   already  filtered  some  junk  traces  which  causing  degrade  of  tracked  path  Statistical  data  of  tracking  error  over  time  (basing  on  real-time  output)  End-to-end  delay  of  each  target  position  data  reaches  BS  Total  residual  energy   of   all   sensor  nodes   and  residual   energy   of   each   sensor   node  at  end  of  simulation  Packet  delivery  rate  of  data  link  layer  (whole  network) IV.3 Result analyses IV.3.1 Tracking accuracy a Tracked trajectories After  target’s  position  is  estimated  at  cluster,  the  CH  relays  this  data  to  BS  in  a   packet   DATA_TO_BS   The   relaying   process   is   repeated   through   multiple   hops   before   the   packet   reaches   BS   The   efficiency   of   network   layer   and   data   link   layer   does   not   directly   affect   tracking   accuracy   at   this   stage   (because   the   estimation   is   carried   out   at   CH);;   however,   it   does   affect   number   of   DATA_TO_BS   packets   delivered  to  BS  Number  of  delivered  DATA_TO_BS  packets  determines  how  detail   the  positioning  data  produced  by  CH  (raw  positioning  data)  that  the  BS  receives  The   detail  of  raw  positioning  data  will  affect  the  quality  of  final  output  (filtered  tracking   path)  produced   by   BS   Figure   IV-2   and   Figure   IV-3   illustrate  raw  positioning   data   delivered  by  two  systems  with  XT-MAC  and  B-MAC  installed  In  the  figures,  each   target  position  vector  is  illustrated  as  a  point  in  three-dimensional  space,  representing   its  three  components  (x,  y,  t),  each  point  is  also  colored  to  represent  its  timestamp  t   more  clearly As  illustrated  in  the  figures,  this  raw  data  is  still  noisy  and  contains  many  jump   points   These   jump   points   degrade   the   tracking   quality   more   than   improve   it   and   should  be  filtered  out  Compare  Figure  IV-2  and  Figure  IV-3,  we  can  easily  see  that   the  data  delivered  by  XT-MAC  is  more  detail  than  the  data  delivered  by  B-MAC  It   means  that  XT-MAC  is  more  reliable  when  relaying  packets  to  BS This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 36 Figure IV-2: Raw positioning data delivered with XT-MAC Figure IV-3: Raw positioning data delivered with B-MAC After  the  raw  positioning  data  is  delivered  to  BS,  the  BS  does  not  use  that  data   directly  to  construct  the  output  (traces)  Instead,  it  use  a  filtering  algorithm  to  process   this   data   and   improve   the   accuracy   In   parallel,   a   post-processing   routine   is   also   executed  to  remove  junk  traces  (containing  unreliable  segments  of  tracked  path  and   jump   points)  The   result   of   post-processing   appears   with   a   small   constant   lag   after   raw  positioning  data  arrival This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 37 Figure  IV-4  and  Figure  IV-5  illustrate  filtered  traces  with  junk  traces  removed   The  detail  of  raw  positioning  data  does  a  great  impact  with  quality  of  filtered  traces   Unreliable  segments  of  tracked  path  are  removed  and  the  remaining  parts  are  more   accurate Figure IV-4: Filtered traces with XT-MAC Figure IV-5: Filtered traces with B-MAC Figure  IV-6  illustrates  the  final  tracking  result  compared  to  true  movement  path   of  the  target  As  you  can  see,  the  proposed  system  design  can  provide  tracking  result   following  closely  to  the  true  path This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 38 Figure IV-6: Tracked path (without timestamps) b Tracing error To  inspect  tracking  quality  produced  by  CH  more  closely,  the  positioning  errors   are  recorded  at  CH  and  plotted  in  the  Figure  IV-7  Tracking  errors  are  also  recorded   at   BS  and  plotted   in   Figure   IV-8   Note   that   tracking  errors  are  recorded  right   after   target  position  vectors  arrive  and  therefore  errors  of  jump  points  are  also  included  in   the   statistics   Compare   these   two   figures   more   closely   we   can   recognize   some   plotted   errors   (with   high   values)   are   unchanged   while   almost   others   are   reduced   thank  to  filtering  process  Because  these  are  jump  points  and  they  are  classified  into   junk  traces  where  number  of  target  position  vectors  is  only  one,  these  traces  are  not   improved  (due  to  lack  of  historical  data)  and  will  be  removed  by  post-processing This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 39 Figure IV-7: Positioning error produced at CH Figure IV-8: Tracking error produced at BS (including jump points) This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 40 Table  IV-2  contains  statistics  about  tracking  errors  It  shows  that  the  mean  error   of  final  output  of  proposed  system  with  XT-MAC  is  4.71  m  and  with  B-MAC  is  7.03   m   In   compare   with   sensing   range   of   35   m,   the   error   of   proposed   system   is   about   13% With XT-MAC With B-MAC Error at CH BS CH BS 6.67 4.71 8.95 7.03 Mean error (m) 6.18 4.29 6.55 5.56 Standard deviation Table IV-2: Tracking errors statistics IV.3.2 End-to-end delay In   order   to   assure   real-time   property,   the   tasks   of   target   detection,   target   localization,   and   target   state   report   need   to   be   completed   within   each   sampling   interval  In  other  word,  the  end-to-end  delay  needs  to  be  smaller  than  the  sampling   interval  (0.5  s)  Note  that  B-MAC  and  XT-MAC  are  configured  differently  because   their  duty  cycling  mechanisms  are  different,  method  for  listening  to  preamble  packet   of   B-MAC   is   based   on   its   own   clear   channel   assessment   mechanism   [1];;   whereas   XT-MAC  needs  to  receives  a  complete  strobe  packet  However,  the  inspected  data  in   this  thesis  are  from  the  configurations  providing  best  result  of  the  two  protocols Because   both   simulated   systems   use   the   same   upper   layers,   they   will   have   similar  traffic  load  In  addition,  both  B-MAC  and  XT-MAC  are  based  on  CSMA/CA   algorithm  for   multiple-access  channel,   the   difference   in   end-to-end  delay   is   mainly   caused  by  duty  cycling  activities  at  each  hop  We  call  the  additional  delay  caused  by   duty   cycling   activities   is   duty   cycling   delay  As   shown   in   Figure   IV-9,   end-to-end   delay   of   most   packets   transmitted   with   XT-MAC   is   smaller   than   0.5   s   or   the   realtime  property  of  the  system  is  assured  B-MAC  provides  fairy  good  results;;  however,   its  efficiency   reduced   significantly   while   number   of   hops   increase   and   most   of   the   time   B-MAC   has   higher   delay   than   XT-MAC   The   average   value   of   end-to-end   delay   of   XT-MAC   is   0.281   s   while   B-MAC's   is   0.453   s   In   details,   two   protocols   send   their   preamble   packet(s)   to   wake   up   receiver(s)   before   sending   each   payload   packet;;   B-MAC   have   to   send   a   long   preamble   packet   (which   is   matched   with   checkInterval  of  receivers),  this  step  is  repeated  at  every  hop  in  relay  path  from  CH   to   BS;;   in   other   words,   at   each   hop,   B-MAC   has   additional   delay   equals   checkInterval  Whereas,   XT-MAC   divides   the   preamble   into   small   strobes   and   has   windows   of   free   channel   between   strobes,   this   enables   receiver   to   send   back   an   acknowledgement  as  soon  as  it  wakes  up  With  EMRP  routing  protocol,  we  usually   have   a   stable   relay   path  for   several   sensing  cycles  With   the  stable   relay   path,   XTMAC  improves  the  delay  even  more,  because  after  being  woken  up  for  the  first  time,   receiver   stays   in   ACTIVE   state   for   a   while   and   it   can   response   immediately   to   strobes  of  next  sensing  cycles This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 41 Figure IV-9: End-to-end delay With XT-MAC With B-MAC 0.281 0.453 Mean (s) 0.1596 0.1388 Standard deviation Table IV-3: End-to-end delay statistics IV.3.3 Energy consumption Energy  consumption  is  also  one  important  aspect  of  WSN  As  shown  in  Figure   IV-10,   after   more   than   300   seconds   of   the   simulation,   the   total   residual   energy   is   equal  to  1031.45  mWh  with  XT-MAC  and  992.4  mWh  with  B-MAC  (initial  value  is   1280   mWh)   Both  systems   achieve   low   power   consumption   in   high  load   condition   and  XT-MAC  performs  slightly  better  than  B-MAC  However,  by  observing  energy   maps   in   Figure   IV-11,   consumed   energy   converges   at   the   nodes   near   BS   more   clearly   with   B-MAC   This   means   that   for   longer   work,   there   will   be   significant   difference  in  network  lifetime  between  two  corresponding  systems This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 42 Figure IV-10: Total residual energy (a) With XT-MAC (b) With B-MAC Figure IV-11: Residual energy maps at end of simulation IV.3.4 Packet delivery rate of data link layer Table  IV-4  is  statistics  about  payload  packet  loss  caused  by  data  link  layer  This   includes  all  packets  from  upper  layers  (network  layer  and  higher)  As  indicated,  both   systems  have  acceptable  packet  loss Lost packets Delivered Packet loss packets percent 3964 36596 9.7% With XT-MAC 4300 33775 11.3% With B-MAC Table IV-4: Packet loss caused by data link layer This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 43 However,  Table  IV-4  only  reflects  general  reliability  of  the  data  link  layer,  this   statistics   is   dominated   by   control   packets   of   upper   layers   and   packets   transmitted   within  clusters  To  examine  more  clearly  about  the  influence  of  these  MAC  protocols   to  tracking  quality  produced  at  BS,  relayed  packets  (DATA_TO_BS)  created  at  CHs   and  delivered  to  BS  are  counted  Counting  result  is  represented  in  the  Table  IV-5 Relayed packets Created Delivered 602 487 In system with XT-MAC 473 337 In system with B-MAC Table IV-5: Delivery rate of relayed packets Delivery rate 81% 71% Easily  see  that  there  are  significant  differences  in  both  number  of  created  relayed   packets  and  their  delivery  rate  The  system  with  B-MAC  has  lower  number  created   relay  packets;;  it  means  that  there  is  more  interference  in  clusters  than  in  the  system   with   XT-MAC,   causing   loss   of   measurement   packets   and   degrading   positioning   quality  of  CHs  In  other  hand,  with  better  delivery  rate,  XT-MAC  helps  to  provide   more   detail   positioning   data   to   BS   and   therefore,   improve   final   tracking   quality   of   BS This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 44 CONCLUSION  AND  FUTURE  DEVELOPMENT In  this  thesis,  overviews  about  wireless  sensor  networks  and  general  structure  of   a   tracking   wireless   sensor   network   have   been   introduced   A   design   for   an   energyefficient,  accurate  and  real-time  target  tracking  system  is  proposed  In  addition,  XTMAC   –   a   duty   cycling   protocol   is   also   proposed   with   features   tailored   for   the   tracking  system An   extensive   simulation   has   been   developed   to   evaluate   the   efficiency   of   the   designed   system,   especially   the   efficiency   contributed   by   the   proposed   MAC   protocol   In   the   simulation,   B-MAC   –   a   classical   duty   cycling   protocol   has   been   implemented  as  a  reference  Simulation  results  confirmed  that  the  new  XT-MAC  can   track   targets   in   real-time   with   reasonable   accuracy   while   achieving   better   energy   consumption   and   delay   compared   to   the   long   preamble   duty   cycling   protocol   BMAC However,  the  proposed  system  still  has  some  deficiencies;;  especially  one  major   limitation  of  the  system  is  that  it  still  lacks  of  ability  to  keep  accurate  tracking  when   two   or   more   targets   move   close   to   each   other   This   limitation   is   caused   by   some   factors  such  as  signal  processing  ability  of  sensing  devices  and  the  accuracy  of  data   association   mechanism   In   the  future,   more   researches   will  be   carried   out   to  attack   these  limitations  of  the  system  and  improve  its  performance This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 45 REFERENCES [1] Polastre, Joseph and Hill, Jason and Culler, David, "Versatile Low Power Media Access for Wireless Sensor Networks," in Proceedings of the 2Nd International Conference on Embedded Networked Sensor Systems, 2004 [2] L Mingxi, X Yan, C Yi, and S Hu, "A mac protocol for target-tracking in wireless sensor network," Chinese Journal of Electronics, vol 22, p 359, 2013 [3] L Song and D Hatzinakos, "A cross-layer architecture of wireless sensor networks for target tracking," IEEE/ACM Transactions on Networking, vol 15, pp 145-158, 2007 [4] Low-Rate Wireless Personal Area Networks (LR-WPANs), New York: IEEE, 2011 [5] Tian He, Sudha Krishnamurthy, Liqian Luo, Ting Yan, Lin Gu, Radu Stoleru, Gang Zhou, Qing Cao, Pascal Vicaire, John A Stankovic, Tarek F Abdelzaher, Jonathan Hui, and Bruce Krogh, "VigilNet: An integrated sensor network system for energy-efficient surveillance," ACM Trans Sen Netw., vol 2, pp 1-38, 2006 [6] Luis Javier García Villalba, Ana Lucila Sandoval Orozco, Alicia Triviño Cabrera and Cláudia Jacy Barenco Abbas, "Routing Protocols in Wireless Sensor Networks," Sensors, vol 9, no 11, pp 8399-8421, 2009 [7] Braginsky, David and Estrin, Deborah, "Rumor Routing Algorthim for Sensor Networks," in Proceedings of the 1st ACM International Workshop on Wireless Sensor Networks and Applications, Atlanta, 2002 [8] Akkaya, K and Younis, M., "An energy-aware QoS routing protocol for wireless sensor networks," in Distributed Computing Systems Workshops, 2003 Proceedings 23rd International Conference on, 2003 [9] Singh, Shio Kumar; Singh, M P.; Singh, D K., "A Survey of Energy-Efficient Hierarchical Cluster-Based Routing in Wireless Sensor Networks," International Journal of Advanced Networking & Applications, vol 2, no 2, pp 570-580, 2010 [10] Bhattacharyya, Debnath and Kim, Tai-hoon and Pal, Subhajit, "A Comparative Study of Wireless Sensor Networks and Their Routing Protocols," Sensors, This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 46 vol 10, pp 10506-10523, 2010 [11] Al-Karaki, J.N and Kamal, A.E., "Routing techniques in wireless sensor networks: a survey," Wireless Communications, IEEE, vol 11, pp 6-28, 2004 [12] Vinh Tran-Quang and Hung Nguyen-Khanh and Thu Ngo-Quynh, "Target tracking system using lateration estimation method in wireless sensor networks," in Ubiquitous and Future Networks (ICUFN), 2013 Fifth International Conference on, 2013 [13] E Brookner, Tracking and Kalman Filtering Made Easy, JOHN WILEY & SONS, INC, 1998 [14] Thuy Tran Vinh, Thu Ngo Quynh, Mai Banh Thi Quynh, "Emrp: Energyaware mesh routing protocol for wireless sensor networks," in Advanced Technologies for Communications (ATC), 2012 International Conference on, 2012 [15] Chih-fan Hsin and Mingyan Liu, "Network coverage using low duty-cycled sensors: Random & coordinated sleep algorithms," in Information Processing in Sensor Networks, 2004 IPSN 2004 Third International Symposium on, 2004 [16] R E Kalman, "A New Approach to Linear Filtering and Prediction Problems," Transactions of the ASME – Journal of Basic Engineering, pp 35-45, 1960 [17] Greg Welch, Gary Bishop, An Introduction to the Kalman Filter, Chapel Hill, 2006 [18] Arulampalam, M.S and Maskell, S and Gordon, N and Clapp, T., "A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking," Signal Processing, IEEE Transactions on, vol 50, no 2, pp 174-188, 2002 [19] A Varga, "The OMNeT++ discrete event simulation system," in EMS, 2001 This thesis is performed by: Nguyen Trung Quan – 20092138 – ICT-541 47 ... Nguyen Trung Quan – 20092138 – ICT-541 iv TÓM  TẮT  NỘI  DUNG  ĐỒ  ÁN  TỐT  NGHIỆP Mạng cảm   biến không dây  (WSN)  là  một   hướng  nghiên  cứu  mới,  mở  ra  cơ  hội   phát   triển  các...   nối không dây  và  hoạt  động  một  cách  độc  lập  trong  môi  trường  mở  và  thường  có  ít  hoặc   không  có  sự  bảo  trì  của  con  người  Chúng  được  gọi  là  các  nút cảm biến  và... Nội  dung  của  đồ  án  chỉ  tập  trung  nghiên  cứu  phát  triển  một  hệ  thống mạng cảm biến   không dây  phục  vụ  cho  ứng  dụng  theo  dõi  mục  tiêu  Chức  năng  chính  của  hệ  thống