Predictive data push technology framework - wireless ... - Science Direct

Copyright© IFAC Programmable Devices and Embedded Systems Bmo, Czech Republic, 2006

" IFAC"

'-U--> Publications

PREDICTIVE DATA PUSH TECHNOLOGY FRAMEWORKWIRELESS USER LOCALIZATION USABILITY IN CONTROL SYSTEMS

Ondrcj Krc.icar, Jindrich Ccrnohorsky

VSB Tcclmica/ University ojOstral'a, FE/, iJepartment of.Heasurement and COlliro/, Centre jor Applied Cyhemetics, 17, lisiopadu /5, 70833 Ostrava-Poruba, Czech Republic ondrej. krejcadii)l'sh. cz, jilldrich. cernohorsky(ii}l'sh. cz

Abstract: The proliferation of mobile computing devices and local-area wireless networks has fostered a growing interest in location-aware systems and services. Additionally, the ability to Ict a mobilc dcvicc dctcnninc its location in an indoor cnvironmcnt at a fincgrained level supports the creation of a new range of mobile control system applications. Main arca of interest is in model of radio-lrequency (RF) based systcm enhancement for locating and tracking users of our control system inside buildings. The framework described here joins the concepts of location and user tracking in an cxtcndcd existing control system. The experimental framework prototype uses a WiFi network infrastructure to let a mobile device determine its indoor position as well as to deliver IP connectivity. User location is used to data pre-buffering and pushing infonnation from server to user's PD/\. Experiments show that location detennination can be realized with a room level granularity. CopyriKhi © 2006 IFAC Keywords: P])PT framework, Location-aware services, Mobile device, .NET Cl", OpenNETCF. control system.

mobile environment. We believe that an important paradigm is contcxt-awareness. Contcxt is relcvant to the mobile user, because in a mobile environment the context is oftcn very dynamic and the user interacts dilTcrcntly with the applications on his mobile device when the context is different. While a desktop machine usually is in a fixed context, a mobile device goes from work , to on the road, to work in-amccting, to homc, ctc . Contcxt is not Iimitcd to thc physical world around the user, but also incorporates the user's behaviour. and terminal and network characteristics.

1. INTRODUCTION Thc usage of various wirelcss tcchnologics that enable convenient continuous lP-level (packet switched) connectivity for mobile devices has increast:d dramatically and will continut; to do so for the coming years. This wi\1 lead to the rise of new application domains each with their own spccilic features and needs. Also. these new domains will undoubtedly apply and reuse existing (software) paradigms, components and applications. Today, this is easily recognized in the miniaturized applications on network-connected P])As that provide more or less the same functionality as their desktop application equivalcnts. Thc web browser application is such an example of reuse. Next to this, it is very likely that these new mobile application domains adapt new paradigms that specifically target the

Context-awareness concepts can be lound as basic principles in long-tcrm strategic rcscarch lor mobile and wireless systems such as formulated in (WWRF , 2(05). The majority of contcxt-aware computing to date has been restricted to location-aware computing

378

for mobile applications (location-based services). However. position or location infomlation is a relatively simple form of contextual information. To name a few other indicators of context awareness that make up the parametric context space: identity, spatial information (location, speed), environmental information (temperature), resources that are nearby (accessible devices, hosts), availability of resources (battery, display, network, bandwidth). physiological measurements (blood pressure, hart rate), activity (walking. running). schedules and agenda settings. Context-awareness means that one is able to use context information. We consider location as prime form of context information. Our focus here is on position dctcrmination in an indoor cnvironmcnt. This indicator is basic level of the predictive data push technology (PDPT) framework described in this paper. Location information is used to determine an actual user position and his future position. When the user position and user track is known, the data which could be needed by the user in the future is pushed to his mobile device . This tcehnique is suitable for data rate increasing between server and user side of cxisting control systcm. whcn thc I'DI'T framcwork is put on. Although the framework functionality is useful by itself, it is specifically targeted at being an environment supporting rich mobile applications in the area of control systems computing. Other systems also combine one or more context elements and incorporate this in a mobilc application. Wc believe, however. that our framework provides a unique combination: the predictive data push technology and location context information as part of an (existing) control system together with an indoor position determination mechanism based on WiFi . The framework ' s location determination mechanism focuses on applications that need to be able to obtain the user' s indoor position with a room-level granularity. We build the framework based on various existing components and technologies and on our own additions and extensions. We have performed a number of experiments with the control system, focusing on position determination, and arc encouraged by the results. The remainder of this paper describes the conceptual and technical details ofthe framework. In scction 2. wc discuss thc cnvironmcnt in which the system runs, focusing on basic concepts and technologies. In scction 3, wc addrcss the framework architecture, followed by the implementation in section 4 . Section 5 describes the experiments and conclusions in section 6. 2. BASTC SYSTEM CONCEPTS AND TECIINOLOGIES The proliferation of mobile computing devices and local-area wireless networks has fostered a growing

interest in location-aware systems and services. A key distinguishing feature of such systems is that the application information and/or interface presented to thc uscr is. in gcneral , a function of his physical location. The granularity of location information needed could vary from one application to another. For example. locating a nearby printer requires fairly coarse-grained location information whereas locating a book in a library would require fine-grained information. While much research has been IOcused on dcvelopmcnt of scrviccs architccturcs for locationaware systems, less attention has been paid to the fundamental and challenging problem of locating and tracking mobile users, especially in in-building cnvironmcnts. Wc focus mainly on RF wireless networks in our research. Our goal is to complement the data networking capabilities of RF wireless LANs with accurate user location and tracking capabilities for user needed data pre- butlering. This property we use as information ground for our control system.

2.1 Location-based services Location-based services (LBS) are touted as 'killer apps' for mobile systems. An important difference between lixed and mobile systems is that the latter operate in a particular context, and may behave dilTcrcntly or olrcr dillerent inlormation and interaction possibilities depending on this context. Location is ol1en the principal aspect determining the context. Many different technologies are used to provide location information. Very common is the GPS system. which uses a network of satellites and provide s position inlormation accurate within 10-20 m. However. due to its satellite based nature. it is not In cellular suited for indoor positioning. telecommunication networks such as GSM, the cell ID gives coarse-grained position information with an accuracy of about 200 m to to km. Morc advanced technologies based on triangulation measure e.g. the angle of arrival (AOA). the time of arrival (TOA). or the enhanced observed time difference (I:-OTD), and typically provide an accuracy of < 150 m. For finc-graincd indoor location information, various technologies are available, based on infrared. RF, or ultrasonic technologies often using some type of beacon or active badge. Given the ubiquity of mobile devices like PDAs. however, active badges will probably be superseded by location technologies incorporated in these devices. In the context of our experimental setup. we need indoor position information accurate enough to determine the room in which the user is located. We must dcploy a scparatc location tcchnology , where we use the information available from a WiFi network infrastructure to determine the location with

379

room-level accuracy. By this information possible u~er tra~k i~ e~timate.

2.2 WiF'i

-It;t..~·

802.11

strength of s Watts and a noise power of n Watts yields an SNR of 1O*loglO(s!n) dB. For example. a signal strength of I Watt is equivalent to 30 dBm. furthermore. if the noise power is 0.1 Watt, the SNR would be 10 dB.

The Institute of Electrical and Electronics Engineers (IEEE) develops and approves standards for a wide variety of computer technologies . IEEE designates networking ~tandard~ with the number 802. Wirde~~ nctworking standards arc dcsignatcd by thc numbcr 11 . IIence. IEEE wireless standards fall under the 802.11 umbrclla . Ethcmct. by the way , is called R023 (Reynolds. 20m).

The WaveLAN driver extracts the SS and the SNR information from the WaveLAN lirmware each time a broadcast packet is received . It then makes the inJormation available to user-level applications via system calls. 11 uses the wlconfig utility, which provides a wrapper around the call s, to extract the signal infonnation.

The 802.11 b is an updated and improved version of the original IEEE 802.11 standard . Most wireless networking products today are based on 802.11 b. 802.11 b networks operate at a maximum speed of II Mbps, slightly faster than IO-BASE-T Ethernet. providing a more than fivcfold incrcasc ovcr thc original 802.11 spec. The 802.11 standard provided for the u~c of OSSS and FHSS spread-spectrum methods. In 802.11 b. OSSS is used.

2.6 Localizatioll melhodology

We use only 802.11 b infrastructure (POA has only thi~ standard) so othcr standards (802.lla or g) is not needed to describe. However. it can be possible to dcvdop a POPT framework with it.

The general principle is that if a Wifi-enabled mobile device is close to ~uch a stationary device Access Point (AP), it can "ask" the location provider's position by setting up a WiFi connection. If the mobile device (pOPT framework) knows the position of the ~tationar)' device . it also know~ that its own position is within a I aa-meter range of this location provider. Granularity of location can improve by triangulation of two or several visi ble WiFi A Ps as descrihed on figure [Fig. I). Thc POPT POA client will ~upport the application in automatically retrieving location information from nearby location providers. and in interacting with the POPT server. Naturally. this principle can be applied to other wireless technolog ies.

2.4 PushillK illlnlrallels

Intranets are simply local 'intcrncts' closed inside various organiLation~. They allow transfer of information between users. Important company news and information can he pushed straight to the Screen of the emplovees. This could be done in a form of screen savers, tickers (scrolling bars of text) or by some client application. A good example of a useful push application within companies is automatic notification of events. for example. by linking a push system to a database of stocks and components in a manufacturing company a manager could be alerted if the number of a particular component drops below a certai n level.

I

/

"

\

/

/

.) i

I I

f

When some intelligence is added to push sy stem. the very useful svstem may he created. This is what we do. Location inlormation about u~ers is added to server as intelligence for push technology.

"".".. _.. _.......... " ... "'~

/1

/ '--_. ,_ _ _ __ ... ~

roorn

2.5 Data Collectioll A key step in the proposed research methodology is the data collection phase. We record information about the radio signal a~ a fun~tion of a user's lo~ation . The signal information is u~cd to construct and validate models for signal propagation.

Operator

Fig. I.

Localization principle - triangulation.

3. THE POPT fRAMEWORK ARCHITECTURE

Among other information. the WaveLAN NlC makes available the signal strength (SS) and the signal-lonoise ratio (SNR). ss is reported in units of dBm and SNR is expressed in dB. A signal strength of s Watts is equivalent to I O*log I O(s/(J.OOI) dRm. A signal

The POPT framework consists of an inlrastructure for handling location context information and an infrastructure for determining indoor position

380

figure [Fig. 2]) record s just one set of signal strength measun:ments. I3y this set of value the actual user position is determined .

information. The POPT framework is primarily Jocu ~t:d on ~upporting n:st:arch and prototypt: mobik applications enhancement in the area of locationaware and control ~ystems computing. For the handling of location and information from control system, the framework provides a PDPT server as well as a client application. The client application is availahle for POAs that run the Microsoft Pocket PC / Windows CE 3.0 operating system with .NET Cl' support.

3.2. Wireless loca/ion architeclllre Another very important part of the project is design of Wireless Location Architecture (WLA). WLA delines a structure lor data store in database. The structure is dclincd as data le vel s in huilding plan for example.

Tht: handling or tht: location inJormation in the framcwork docs not dcpcnd on thc granul arity of thc location determination mechani sm. The framework can handle any kind of location information from any source, although the focus here is on indoor location "'1th a room-level granularity . The indoor position can be retrieved from an infrastructure of stationary WiFi devices that know their own location. A stationary device has location provider characteristics.

The example of function: User location is determined and anal yzcd . Scrvcr activates the PDPT and pushes data to PDA. As tirst the data about Building Block will be copied. As next the Block Floor data and finall y the data about user occurred cell ",ill be copied .

_E!uj~jl'!.9J3tg9clt1,mJ"" ...-: Block AoorLev

3.1. J'r edictive data Pllsh techllology

Cell Level

Thi s part of project is based on model of locationaware enhancement, which we used in created control system. These information about are useful in framework to increase real datafl ow from wireless acct:~ ~ point (st:rvt:r sidt:) to PDA (client side). Primary dataflow is enlarged hy data pre-hutTering. The~ e techniques lorm the basi s lor predicti vc data push technology (POPT). POPT copies data from inJormation st:rver to clients PDA to be on hand when user comes at desired location.

Fig . 3.

WLA lor building.

3.3. Frame work desiKIl POPT framework de sign is based nn most commonl y used server-client architecture . To process data the server has online connection to the control system. Data from technology arc continually saved to SQL Server database (WLA architecture) (Till'any. 2003) and (Reynolds , 2003 ). The part of thi s database (dcsired b,· uscr locati on or hi s demand ) is replicated online to client's POA where it is vi sualized on the screen. User P])A has location sen sor component which continuously send s to the framework kernel the information about nearby AP' s inten sity. The kernel processes this information and makes a decision if and how a part of WLA SQL Server databa se will be replicated to client's SQL Server CE databa se.

The benefit of PJ)PT consists in reduction of time needed to di splay desired information requested by a user command on POA . Time delay may vary from a few seconds to number or minutes. It depend s on two aspects. First one is the quality of wireless Wi-Fi connection used hy client PJ)A . A theoretic speed of Wi-Fi connection is m3X ~Q5 ill/s. However. the test of transfer rate from server to client' s PDA , which wc have carried out within our Wi-Fi infrastructure provided the result speed only 1(,0 KFl!s. The second aspect is the si ze of copied data.

POPT server ~~Punl.

f"'"" ~SQC2

r "" DSSID

I:__,~ram~work

. . ..

• 0 ·-= 0 "' - = 0 • -===::::J 0

KOTNET lIP )

rllli!lllity

Framework itemel

.1

SigrwrI -67

tlo!lllMec l SU26_m ms'VI

-6B

I: OlN:.i

~

5 01.. eorver

.~

/ ' L--_ _ _ _ _ _ _ _....J

Fig. 2.

Control System

Loeator - AP intensity & Positioning.

The application (locator) based on .NET language is now created lor testing. Current application (see

rig.4.

381

System arehitt:cture - UML dcsign

if

The kernel decisions constitute the most important part of whole framework because the kernel must continually compute the position of the user and track and make a prediction of his future movement. After doing this prediction the appropriate data (part of WLA SQL Server database) are pre-buffered to elient's database for future possible requirements.

(adpt.:. Isr..arc!css)

{

foreach

(Access Point ap in Clop:"'r . Nearby.n.ccessPoi nts 1

dtVisibleAP . NewH::M( ; ; drData Row [ "BSSID"l =( i;:I. f.>.N bme. TcStr'inq (: ) ; drDataHow

=

drData Row( "Sig nu l [% ] "J = (lap . ~ i g n al~t r engt h .lle~ i b els) . TeSt=ing ( )) ; d:V i s i blc!Ik' . l-\o ...'s.f\dd (d r Dat ak ~w) ;

4. THE IMPLEMENTA nON The implementation of the PDPT framework and application consists of a numbcr of componcnts. The POA cl ient location determination software is implcmcnted by the WiFi middlcware.

Fig. 6.

Sample code - signal strength from AP

4.3. Sen'er Basic idea of PDPT framework is in connection between location information and data push technology . When wc know user position and user track_ the data which could be needed by the user in thc futurc is pushcd to his mobilc devicc.

4.1. PDA client The POPT POA client applications provide location and control system functionality to thc mobilc uscr. It is implemented in CII using the Microsoft Visual Studio .NET 2003 with compact framework and a special OpcnNETCF library cnhanccmcnt (Tiffany, 2003) and (OpeNETCF. 2005). OpenNETCF.org is an independent source for Compact Framework development information working under the spirit of the open-source movement.

4.2. WiFi middle ware The WiFi middleware implement s the client' s side of location determination mechanism on the Windows CE 3.0 PockctPC operating system and is part of the POPT POA client application. The libraries used to manage WiFi middlcwarc arc li sted on figure rFig . 5).

5. EXl'ERlMENTS We have executed a number of indoor experiments with the PDPT framework , using the PDPT PDA application . WiFi access points arc placed at different locations in the building, where the access point cells partly overlap. We used triangulation principle of AP intensity to get better granularity. We found that the location determination mechanism selects the access point that is closest to the mobile user as the best location provider. Also, after the loss of IP connectivity, the switch from one access point to another (a new best location provider) takes place within a second in the majority of cases. resulting in only tcmporary loss of 11' conncctivity. This technique uses a special Radius server (RADnrS , 200S) to realize roaming known in cell networks.

Adapter AecessPoint Networking SignalSlrenglh AccessPointCollcction AdapterCollection AdapterType ConncetionStatus NctworkType

Open NETCF .Net.

Fig. S.

The server application GUT uses the Microsoft Foundation Classes (MFC) with .NET platform. Framework server manages MS SQL Server 2000 with a special WLA architecture. The special search algorithm is implemented in framework kernel to manage user location information used for hetter server response. This algorithm is the most important part of the framework and much other development work will be spent on this problem in the futun;.

Currently. the usability of the PDPT PDA application is somewhat limited due to the fact that the device has to be continuously powcrcd. If not, the WiFi interface and the application cannot execute the location determination algorithm, and the POPT server docs not rcccivc location updatcs from the PDA client. We have executed a number of tests of battery power consumption with three PDA devices. The tests were executed from 100 % battery level to 20 % hattery level with halanced load.

OpenNETCF network method architecture The first was HTC Blue Angel PH20B which is known also as MDA TIT from T-Mobile CTntel XScale PXA263 CPU, MS WM2005 OS). The second onc was iPAQ h4150 from Hewlett & Packard Company (lI&P) (lnte! XScale PXA255 CPU, MS WM2003 OS). The third one was iPAQ hx4700 from H&P (lntel XSeale PXA270 CPU, MS WM2003 SF. OS).

Methods from the Net library are used for example to display Visible WiFi AP. See figure [Fig. 6). dlVl!..i.ii:..\':'~ .n.['

=

C