Presentation is loading. Please wait.

Presentation is loading. Please wait.

Localization in Wireless LANs. Outline  Wireless LAN fundamentals  Wi-Fi Scanner  WLAN Localization  Simple Point Matching  Area Based Probability.

Similar presentations


Presentation on theme: "Localization in Wireless LANs. Outline  Wireless LAN fundamentals  Wi-Fi Scanner  WLAN Localization  Simple Point Matching  Area Based Probability."— Presentation transcript:

1 Localization in Wireless LANs

2 Outline  Wireless LAN fundamentals  Wi-Fi Scanner  WLAN Localization  Simple Point Matching  Area Based Probability

3 Wireless LAN  Standard  IEEE 802.11a  IEEE 802.11b  Also call “Wi-Fi”  operating at 2.4 GHz  11 Mbps  IEEE 802.11g  operating at 2.4 GHz  54 Mbps  Future Standard

4 Wireless LAN  Information  MAC Address  Identifier of the Wireless LAN Access Point (AP)  Provided by the Ethernet LAN in the AP  RSSI  Signal Strength  SSID  Name of AP

5 Wi-Fi Scanner  Platform  Pocket PC 2003(Windows CE 4.0)  Wi-Fi Network  IEEE 802.11b  API  Windows CE.NET 4.2  Tools  Embedded Visual C++ 4.0  Visual Studio.NET 2003

6 Wi-Fi Scanner  Target  Get two unique information  MAC Address  Signal Strength  Future Application  Develop the 2D Location Algorithm  Provide the Multimedia Services (e.g. streaming service)

7 Wi-Fi Scanner   Overall Architecture User InterfcaeCore Operations Embedded VC++ DLL (dynamic link library ) Visual.NET 2003

8 Wi-Fi Scanner Wi-Fi Application MAC Address, Signal Strength Network Driver Interface Specification (NDIS) Application Presentation Session Transport Network Data Link Physical

9 NDIS  Develop the Network Driver  Support varieties of Network Technology (e.g. Ethernet (IEEE 802.3), Token Ring (IEEE 802.5), and IrDA media)  Portability of drivers between platforms that support NDIS  Network adapter miniport driver

10 NDIS Architecture Miniport driver -communicate directly with network interface card (NIC)

11 NDIS User-mode I/O(NDISUIO)  Supports sending and receiving Ethernet frames  Retrieve MediaSense indications  Retrieve signal power indications

12 NDIS User-mode I/O  Steps to bind the NIC Card  Get Ethernet LAN Adapter (NIC) Name  Create the File Handle to bind NDISUIO  Using the “DeviceIoControl” interface to achieve the required packet

13 NDIS User-mode I/O  fRetVal = DeviceIoControl( hNdisUio, hNdisUio,IOCTL_NDISUIO_QUERY_OID_VALUE, (LPVOID) pQueryOid, dwQueryBufferSize, dwQueryBufferSize,&dwBytesReturned,NULL); Retrieves NDIS object

14 NDIS User-mode I/O  struct _NDIS_WLAN_BSSID { ULONG Length; NDIS_802_11_MAC_ADDRESS MacAddress; Uchar Reserved[2]; NDIS_802_11_SSID Ssid; ULONG Privacy; NDIS_802_11_RSSI Rssi; NDIS_802_11_NETWORK_TYPE NetworkTypeInUse; NDIS_802_11_CONFIGURATION Configuration; NDIS_802_11_NETWORK_INFRASTRUCTURE InfrastructureMode; NDIS_802_11_RATES SupportedRates; } NDIS_WLAN_BSSID, *PNDIS_WLAN_BSSID;

15 WLAN Localization  Point-based approach  Localization goal is to return a single point for the mobile object  Area-based approach  Localization goal is to return the possible locations of the mobile object as an area rather than a single point

16 Area-based Approach

17 Advantage of Area-based Approach  Direct the user in the search for an object in a more systematic manner  Presents the user an understanding of the system in a more natural and intuitive manner

18 Some Terms and Definitions  n Access Points  AP 1, AP 2, …, AP n  Training set T 0  the offline measured signal strengths and locations an algorithm uses  Consists of a set of fingerprints (S i ) at m different areas A i  T 0 = ( A i, S i ), i = 1 … m

19 Some Terms and Definitions  Fingerprints S i  Set of n signal strengths at A i, one per each access point  The are totally n access points  S i = (s i1, …, s in ), where s ij is the expected average signal strength from AP j

20 Generating Training Set  In one particular A i, we read a series of signal strengths (s ijk ) for a particular AP j with a constant time between samples  k = 1… o ij,where o ij is the number of samples from AP j at A i  We estimate s ij by averaging the series, {s ij1, s ij2 …, s ijo }

21 Generating Training Set  We do the same for all n APs, so we have the fingerprints at A i,  S i = (s i1, …, s in )  We do the same for all m areas, so we have the training set  T 0 = ( A i, S i ), i = 1… m

22 Getting Testing Set  The object to be localized collects a set of received signal strengths (RSS) when it is at certain location  A testing set(S t ) is created similar to the fingerprints in the training set  It is a set of average signal strengths from n APs, S t = (s t1, …, s tn )

23 Area-based Approach Algorithms  How to use the training set and testing set?  Simple Point Matching  Area Based Probability

24 Simple Point Matching  Compare the received signal strength (RRS) in the training set and the testing set  Find n set of areas that fall within a threshold of the RSS for each AP j, j =1…n  The RSS with threshold for AP j at position i = s ij ±q  Return the areas formed by intersecting all matched areas from different AP area sets

25 Simple Point Matching  How to choose the threshold?  q is the standard deviation of signal received from the corresponding AP  The algorithm starts with a very small q  Area sets for some AP may be empty  q is additively increases eg. q, 2q, 3q …

26 SPM algorithm

27 Area Based Probability  Goal is to return the area with the highest probability that the object is in  Approach is to compute the likelihood of the testing set (S t ) that matches the fingerprint for each area (S i )

28 Area Based Probability Assumptions:  Signal received from different APs are independent  For each AP j, j = 1…n, the sequence of RSS s ijk, k = 1… o ij, at each A i in T o is modeled as a Gaussian distribution

29 Bayes’ rule  We compute the probability of being at different areas A i, on given the testing set S t  P(A i |S t ) = P(S t |A i )* P(A i )/ P(S t ) (1)  P(S t ) is a constant  Assume the object is equally likely to be at any location. P(A i ) is a constant  P(A i |S t ) = c*P(S t |A i )(2)

30 Area Based Probability  We compute P(S t |A i ) for every area A i,i=1…m,using the Gaussian assumption  Max{P(A i |S t ) } = Max{c*P(S t |A i ) } = Max{P(S t |A i ) } = Max{P(S t |A i ) }  Return the area A i with top probability

31 Area Based Probability

32 Reference   Eiman Elnahrawy, Xiaoyan Li, Richard P. Martin,Using Area-based Presentations and Metrics for Localization Systems in Wireless LANs, Department of Computer Science, Rutgers University   Andreas Haeberlen, Eliot Flannery, Andrew M. Ladd, Algis Rudys, Dan S. Wallach and Lydia E. Kavraki, Practical Robust Localization over Large-Scale 802.11 Wireless Networks, Rice University


Download ppt "Localization in Wireless LANs. Outline  Wireless LAN fundamentals  Wi-Fi Scanner  WLAN Localization  Simple Point Matching  Area Based Probability."

Similar presentations


Ads by Google