CesareAlippi,GiovanniVanini

1
A RSSI based and calibrated centralized
localization technique for Wireless Sensor
Networks ????????????????????????

• CesareAlippi,GiovanniVanini
• DipartimentodiElettronicaeInformazione
• PolitecnicodiMilano
• P.zaL.daVinci32,20133Milano,Italy
• Emailalippivanini_at_elet.polimi.it

2
Abstract

• This paper presents a multi-hop localization
  technique For WSNs exploiting acquired received
  signals trength indications.The proposed system
  aims at providing an effective solution for
  theselflocalization of nodes in
  static/semi-Static wireless sensor networks
  without requiring previous Deployment
  information.
• ????????????????????????????????????????????/?????
  ????????????????????????????????????

3
1?Introduction(??)

• In this paper we propose a practical approach for
  units localization in a typical scenario where
  some nodes, e.g., the one sat the network border,
  are used as anchors (i.e. they are placed in
  known positions ). All the RSSI values of the
  packets exchanged among nodes at different power
  levels are collected (RF mapping phase) and
  processed both to build the ranging model to be
  fed into a centralized Minimum Least Square (MLS)
  algorithm. The ranging model is calibrated from
  the online collected values, by selecting the
  optimal approximation family on the specific
  environment And normalizing the intensity of the
  received power.
• ??????????????????????????????????????????????(
  ?????????),???????????????????????rssi?????,??????
  ???????????????????????????family?????????????????
  ???rssi??????

4

• Our contribution can be summarized as
• a practical self localization system which does
  not Need the development of costly deployments of
  thenodes
• a ranging model derived from calibrated RSS
  Information.
• ?????????
• ??????????????????????
• ??????rssi??????

5
2?Localization scenario(????)

• 1.RF mapping of the network it is obtained by
  conveying short packets at different power levels
  through the network and by storing the average
  RSSI value of the Received packets in memory
  tables (we suggest to consider 5-10 values for
  estimating the RSSI )
• 2.creation of the ranging model all the tuples
  recorded Between two anchors are processed at the
  central Unit to compensate the nonlinearity and
  calibrate the model.
• 3.centralized localization algorithm an
  optimization Problem is solved and provides the
  position of the nodes.
• ???????????????
• ???RF????????????????????????????????????rssi????
  (??????5-10????rssi??)
• ???????????????????????????????????????????
• ???????????????????????

6
3?Generation of the RF attenuation modelRF
???????

• 1.the approximation function family. While in
  indoor Environments developing an attenuation
  model is extremely difficult and we have to rely
  on empirical models based created on data
  campaigns or heavy coverage of the environment
  with anchors (which implies a costly deployment),
  outdoor measurements are proportional to
  1/r3-1/r4 ,where r is the distance From the
  emitting and receiving units. The model depends
  on the environment and the distance of the units
  From the soil.
• ??????rf???? ,????????
• 1?????family?????????????????????,???????????????
  ??????????????????(???????),??????1/r3-1/r4???,??r
  ??????????????????????????????????

7

• 2.the calibration over the specie environment,
  which Takes into account the specie ambient
  response.
• 3.uctuation associated with the hardware which
  Changes due to production process. As such nodes
  are different.
• 4.the number of measurements to be considered to
  reduce the influence of the noise.
• 5.the number of information used to configure the
  model
• 2????????,?????????????
• 3??????????????????
• 4??????????????
• 5????????????

8

• Our final approach has similarities to the one
  presented In 8 but we differentiate by
  investigating different model families,
  identified through the experimental campaign.
  Furthermore we consider all the available link
  information To build the ranging model with
  transmitters modulated at Different power levels.
  This point required us to study the Relationship
  between the transmitted and the received power
  And to properly normalize the data to the same tx
  power. More in detail, we rely on the attenuation
  family.
• PTab/rk
• ??????????8???????,???????????????,???????????????
  ??????????????????????????????????????????????????
  ????????????????????,??????????
• PTab/rk

9
4?Localization algorithm(????)

• The proposed algorithm estimates the node
  positions by Minimizing the sum of the
  discrepancies between the estimated distance
  between the nodes and the measure done (Minimum
  Least Square algorithm).
• ?????????????????????????????????(??2??)?

10

• The first step is to correct the received power
  as
• ???????????????
• Prxf(Prx,Ptx)
• The estimated distance between the nodes is then
• ??????????
• The final solution is obtained by minimizing the
  constrained function
• ???????????????

11
5?Experimental results(????)

• The effectiveness of our localization methodology
  has Been tested on a real scenario of outdoor
  localization20 MICA2s are placed on the ground
  in a football field, to cover an area of about
  500mm.
• ??????????????????????20?mica2???????????,??????5
  00???

12

• According to our previous analysis, the first
  stage is To create the ranging model for the
  specific environment. Some tests have been
  performed with all the available data (not only
  the link between anchors) to see which was the
  effect of the linearity correction (power
  calibration) and of the Different approximation
  functions when building the model
• ???????,???????????????????????????????(?????????)
  ?????(????),???????????????????(???????4??)

13

• The overall average error of the localization
  algorithm is given in Fig.5. Results confirm that
  the family 1/r3 Is the best approximation in our
  environment. The localization results provide an
  error of around 3m with 6 anchors and around 2.3m
  with 7 anchors
• ??????????????5??????????????????1/r3????
  .????????6??????????3?????????????2.3??

14
6?Conclusion(??)

• The paper suggests a RSSI-based centralized
  localization technique for outdoor environments.
  Our methodology allows for addressing any outdoor
  environment without complex previous offline
  calibrations and takes advantage of a MLS
  localization algorithm. Experiments confirm The
  effectiveness of this approach which is
  comparable to the best implemented RSSI-based
  multi-hop localization systems we experienced an
  average Error of less than 3 meters when the node
  density is of about One node over 25mm in an area
  of about 500mm .
• ???????????????rssi???????????????????????????????
  ????,??????2?????????????rssi??????,??????????????
  ?500??????????????25??????,??????3????.

15
An Improvement of DV-Hop Algorithm in Wireless
Sensor Networks ????????dv-hop?????

• Wei-Wei Ji and Zhong Liu
• Department of Electronic Engineering
• Nanjing University of Science Technology
• Nanjing, The People!s Republic of China
• jwwlr_at_163.com, eezliu_at_mail.njust.edu.cn

16
Abstract

• In this paper, we develop a new estimation model
  and improve the DV-Hop algorithm by considering
  the relationships between the communication
  ranges and the hop-distances. This scheme needs
  no additional hardware support and can be
  implemented in a distributed way. Simulation
  results show the performance of the proposed
  algorithm is superior to that of the DV-Hop
  algorithm.
• ??,????????????????????????????????Dv-hop?????????
  ????????????????????????????????????????Dv-hop????
  ??

17
1?Introduction(??)

• In this paper, we improved the unknown node
  estimation in the third phase of DV-Hop
  algorithm without additional hardware. In the
  proposed scheme, a constraint is assumed in
• DV-Hop by confining the range from the
  unknown node to reference node in the smallest
  range to the reference nodes. The scheme can be
  implemented in a distributed way and is a
• real range-free algorithm which is
  different from aforementioned algorithms.
• ??,??????????????Dv-hop????????????????????,??
  ??????????????????????????????????????????Dv-hop??
  ???

18
2. DV-Hop Localization Algorithm dv-hop????

• After the first phase, all nodes in the network
  get the minimum hop count values to all reference
  nodes.
• ??????,???????????????????????
• In the second phase, the average single hop
  distance is estimated to convert hop count value
  into physical distance.
• ????,???????????????????????
• In the third phase, the unknown node locations
  can be estimated by the multilateration method
  when these nodes have the distance estimations to
  at least three reference nodes
• ????,???????????3???????????,????????????????????

19
3. Improved Algorithm(????)

• We improve the DV-Hop algorithm by using above
  observation and denote it as CDV-Hop.value
  between N and reference node is L and the
  average single hop distance is HopSizei , i
  1,2, M and M is the number of the reference
  nodes. Then the distance between N and the i-th
  reference node is d LHopSize . Let L
  minL1,L2,LM.In general, there exist several
  reference nodes with same hop values equal to
  L.Denote these reference nodes as vj , j1,2,
  M. Then the location of N can be computed as
• ????CDV-HOP???DV-HOP?????????i,Nu?????i?????Li,???
  ?????HopSizei,i1,2,M,??M????????,i,????????i????
  ????diLiHopSizei,?Lmin L1, L2,
  .LM,????,?????????L????????????????vj,
  j1,2,M,??Nu????????

20
4. Simulation Results(???? )

• Figure 2 is the localization results of UNs by
  the proposed CDV-Hop algorithm and the
  conventional DV-Hop algorithm. In the
  simulations, RNs are set to be 20, UNs are 80 and
  the communication range (CR) is 15. It can be
  seen that the localization accuracy of the
  CDV-Hop algorithm is better than the DV-Hop
  algorithm can do.
• ?2???CDV-HOP?????DV-HOP????????????,??????20,?????
  80,??????15???????CDV-HOP???????DV-HOP??

21

• Figure 3 shows the variation of the average
  localization errors as the reference ratios.
  Suppose that the total . It can be seen from Fig.
  3 that the average localization error by the
  CDV-Hop algorithm is obviously less than the
  DV-Hop method in all considered conditions. And
  the average localization error increases when the
  communication range increases as expected, since
  the accuracy of the average single hop HopSize
  is more coarser with increased distance
  communication range.
• ?3????????????????????????3??????????CDV-HOP??????
  ????DV-HOP??????????????????????,?????????????????
  ?HopSizei?????

22

• Figure 4 is the variation of the normalized
  average localization error with respect to the
  communication range. It is clear that the
  normalized average localization errors decrease
  as the communication range increases.
• ?4?????????????????????????????????,???????????

23
6?Conclusion(??)

• In this paper, we proposed an improved DV-Hop
  algorithm for locating the unknown nodes. A new
  estimation model is developed by considering the
  relationships between the communication ranges
  and the hop values. The simulation results show
  that our proposed method can reduce the nodes
  average localization error significantly in
  different communication ranges and reference node
  ratios.
• ????????????DV-HOP????????????????????????????????
  ??????????????????????????????????????????????????
  ?

24
Random Sampling Algorithm in RFID Indoor Location
System RFID????????????

• Bao Xu, Wang Gang
• UWB Lab, Dept. of Telecommunication
  Engineering,Jiangsu University,Zhenjiang, Jiangsu
  
• Province, China, 212013
• mousebaoxu_at_yahoo.com.cn,gwang_at_ujs.edu.cn

25
Abstract

• ?????????RFID??????,??????RFID?????????????????RFI
  D???????????????????????,?????????????????????????
  ?????????????????????????Toa??2???????????????????
  ?????????????2?????????2????????3???,????????????

26
1?Introduction(??)

• ??,???????????RFID????????????????????RFID???????,
  ????????????????????,?????????????????????????????
  ??????????????,???????,???????????????????????????
  ???????????????????????,????????????????4040?????
  ??,????5???????????????????????,??????????????????
  ????????,???????

27
2. Probabilistic RFID reader model (observation
model)

• ?????????????????????,????P(Xk
  Zk),??Xk??K?????????,Zk?K?????????????????????????
  ,?????????Xk???
• ????,????????P(Xk Zk),????????Xk??????Zk?????????
  ???????????????????????????????????????????,??????
  ????????????

28

• ?????????,????????????,?????????????6?,?????0.9???
  ???????0.5?

29
3. Localization with RFID tags

• 1.?????????,?????????? RkmgtRmax.
• 2.??N??Xk-1?????????????Sk??Xk-1
• 3.???????Sk(Pi)????0?1?????????P(Xk Sk-1i)
• 4.?? , ??
  M????????????
• 5.??N??????????????
• 6.????Xk.
• 7.Kk1
• 8.?????

30
4. Simulation results

• ????RFID???????,???4040???????,??5???????????????
  ?????(10,4), (12,7), (14,5), (15,7), (15,9),
  (15,13), (13,15), (11,15), (9,18), (8,20),
  (8,22), (8,25), (9,25), (12,25), (12,28),
  (14,35), (15,38), (14,40).????,?????NLOS??????????
  ????LS??????????????,?????????????????????

31
(No Transcript)
32
(No Transcript)
33

• ?2,4??NLOS5m?NLOS10m?????,?3,5???NLOS5m?NLOS10
  m???????2??????????????????,??LS???????NLOS???????
  ?????,?????NLOS???????2,4??,?????NLOS??????,??????
  ???

34

• ??6,7??????????2?NLOS?????????????????????3?15????
  ?????LS??????6?????,????????????

35

• ??8 ?????NLOS?0???????????????????????250?????????
  ?????,???????????????????????

36
6?Conclusion(??)

• ????,???????RFID??????????????????????????????????
  ?????????????????????,????????????????????????????
  
• ????,??????NLOS???????????????????????????RFID????
  ???,????????????????????