Automated Wireless Identification System (AWIS) For Observing Social Interactions Of Animals In
The Wild:
A Review Of Bluetooth Wireless API’s
Samantha Knights Computer Science Honours
Rhodes University
May 2004
Introduction
Individual animal identification and social interaction monitoring is necessary for tracking animal behaviour patterns. Currently there are a number of automated animal tracking methods – such as GPS, infrared imagery, GSM and satellite.
Tagging animals for individual identification and monitoring, however, requires human transposition. Even transponder tags require some human interaction of observations made by researchers in the field.
This project proposes the use of Bluetooth Wireless Technology to monitor individual animals’ interactions by placing an enabled device (collar) on each animal – creating an Automated Wireless Identification System (AWIS).
These would transmit information as each tagged animal comes into a certain range of another. Historic information stored on the collar would be transferred to a base station or access point as an animal passes it. These could be mounted on trees or rocks and would in turn transmit collected data to a central server.
The following characteristics of Bluetooth provide motivation for its utilisation this application:
• limited range of 10m,
• low power usage,
• low cost,
• compact size.
The Bluetooth range is important as animals with Bluetooth devices will have the ability to come in and out of contact with each other, signifying some social interaction. Low power consumption is vital as it would be impractical and intrusive for game rangers to regularly change and charge devices attached to the animals. A small size is also essential to reduce interference and discomfort of the animals under observation.
This paper examines current animal tracking methods as these are the only automated techniques for examining behaviour. Wireless Networking technology is suggested as a solution and various different technologies are discussed. Bluetooth is discussed in more detail, with specific reference to the
Bluetooth protocol stack. Available Application Programming Interfaces (APIs) are introduced as well as known implementation issues.
Current Animal Tracking Methods
The World Wildlife Fund (WWF) currently uses radio collars to track polar bears movements in their natural habitats (WWF’s Polar Bear Tracker 2004 [1]). The collars are tracked via satellite and used to ascertain the range of travel of individual animals, as well as their adaptations to changing climates. According to Stephenson 2003 [2] elephants are also collared and tracked in national parks in central Africa. Data such as home ranges, movement patterns and habitat usage are collected. This is used for following “elephant migration corridors” and the development of “transboundary anti-poaching” strategies.
Other methods such as FLIR (Forward-Looking Infrared) Imagery were experimented with by Amstrup et al 2004 [3]. Fitting a device to individual polar bears for such a large area is expensive and “unacceptable as routine
management practice”. Instead, a polar bear’s own body heat could be detected with FLIR Imagery technology. Inhabited dens should thus produce a
pronounced “heat signature”. This was successful but only under the correct conditions; bad weather, sunlight (and other major heat sources), depths too great to be detected led to unfavourable results.
An important discussion in Alibhai et al. 2001 [4] suggests that radio collars are not ideal for black rhinos. It emphasised that routine radio-collaring had a number of drawbacks. According to a study of 89 collars: 28 hose collars, 61 strap collars it was found that: they were expensive to replace, often resulted in wounding (15 percent caused injury and 13 percent were lost and never recovered), difficulty in maintenance as expertise and time required when transmitters become faulty. There was also possibility of compromise in female fertility. Within 12 months more than half radio-collars on males, and just less than half on females had failed. The authors suggest that radio-collaring is not a viable long-term strategy for protection and monitoring.
Steiner et al. 2000 [5] attempted to use GPS technology to track homing pigeons and small animals (dogs). A GPS logger was mounted onto the back of
the animal. The accuracy was only limited to 10m. Tracking the dogs was less successful than pigeons, particularly when they entered denser mediums such as forests. The main issues were power and size constraints. It is suggested that these devices could be used to test hypothesis of pigeon homing, spatial behaviour and orientation.
A commercial company in Sweden, TVP Positioning AB, provides details of their various GPS collar products. Their GPS collars allow for remote data download and reprogramming via either GSM, satellite or radio. Weights range from 350g to 1000g minimum. The features of these collars suggest what would be required for any animal collar: a design that minimises impact on the animal (comfort, size), low battery indicator, mortality indicator, memory larger than battery life-time – to avoid data loss also non-volatile memory, temperature tested, tracking schedules and so on.
‘BlueTrak' is a patented automated system using Bluetooth to gather interaction data for both animals in the wild as well as other applications – such as monitoring physical conditions of the elderly in nursing homes. (Song et al.
2002 [36]). Not much technical detail is provided as BlueTrak is a commercial system. Although the design specification seems to match what is being
attempted in this project, the implementation could be different (the technology we are using was not necessarily available at the time of BlueTrak’s
development)..
All other examples of animal tracking suggest a central data sink (host on some central receiver) where all data is sent by multiple sources (these are the animal’s collars). This is one option – another would be peer to peer connections where certain peers would pass collected data to the data sink. These ideas are used in networks. In order for the collars to be effective they would have to transmit data wirelessly. This is the motivation for extending the concept of wireless networks.
Introduction to Bluetooth
Karadach et al 2000 [18] marks the beginnings of Bluetooth and provides examples of “Bluetooth radios in the wild.”
The first question often asked by people about Bluetooth, rather than its technical specifications, is where the name came from. According to Haartsen et al. 1998 [7], the founding members of the Bluetooth Special Interest Group, the codename was created by James Karach of Intel. Apparently inspired by the Danish Viking King Harald Bluetooth (dates vary around 910-986). Franklin 2004 [8]
hypothesizes that Bluetooth’s name suggests the importance of companies in the Baltic region – Denmark, Sweden, Norway and Finland - to the communications industry.
Bluetooth is epitomized as being a standardised platform to eliminate cables and establish connections. Connections are limited to a10 m range with low-power consumption for battery operated mobile devices. It provides voice, data and audio connections (Bluetooth SIG 2004 [9]).
Bluetooth operates in unlicensed 2.4 GHz radio spectrum – allowing for global use and compatibility. To reduce interference the protocol requires frequency hoping, full-duplex – at 1600 hops per sec. These are 79 frequencies spaced at 1Mhz intervals. The data rate quoted as 1Mbits/s, allowing for point-to- point connectivity for up to 7 simultaneous connections. The key elements to consider when comparing Bluetooth to other wireless technologies are power consumption and intended use. ([9])
Other Wireless Technologies
Dursch et al. 2004 [11] suggest 5 other main wireless technologies available:
Induction Wireless, Infrared Wireless, ZigBee, IEEE 802.11b (Wi-Fi). Dividney 2003 [12] considers Bluetooth, Wi-Fi and IrDA been the most popular standards in 2003. Each have a full qualification and specification program “backed by industry organisation.”
Infrared Wireless: could have been first for transfer data over short distance - approximately 1 m at rates of 4 Mbits/s and 16 Mbits/s. ([11]). IrDA is the Infrared Data Association standard for infrared communication (Survak 2000 [13]). The technology is universally adopted, providing a universal hardware port and approaching software interoperability. IrDA is limited to two participants, point-to-point connection and requires line-of-sight to function ([10]). This
makes IrDA it a non-viable option for animal tagging as there could be a variety of obstacles between the animals themselves (trees, boulders, etc).
IEEE 802.11b (Wi-Fi): standard specification for wireless LAN and works on the same frequency band as Bluetooth ([11]). Wi-Fi power
consumption is considerably larger: currently a Bluetooth device’s minimum current requirements are one tenth of Wireless LAN. (Pico Communications 2001 [14]). Thus Wi-Fi’s range is thus a tenth greater than Bluetooth ([12]). Pico Communications [14] points out, in Bluetooth’s defense, that its range can be extended with the concession of greater power consumption. This is possible because Bluetooth devices are divided into three classes based on their range;
Class 1 has a maximum range of 100m, Class 2 range of 15m and Class 3 the smallest – 10m. It would be possible to track up to 60m of lower class devices with one Class 1 device.
Nevertheless Wi-Fi’s “always on connection model” limits its use in
“PDAs, phones, and other lightweight mobile devices.” ([12]). Bluetooth devices are also much smaller and cheaper than Wi-Fi units. ([10]).
Induction Wireless: invented by Aura Communications, a magnetic field is used for data transfer (as opposed to an electric and magnetic field used by Bluetooth and other wireless technology) – [11]. It has maximum radius of 3m and slower speeds than Bluetooth perhaps making it less suitable for animal tracking. However, it has a lower power consumption and cost when compared to Bluetooth.
Other emerging technologies:
Ultra Wideband: is characterized by a high transfer rate using baseband pulses which use high frequencies and extremely low power. Its suggested applications are location positioning and radar ([15]). Because of its transmission over a large range of radio frequencies (3.1-10.6GHz at 100Mbps [1]), UWB will take a number of years to become accepted worldwide ([12]). Bowles 2004 [16]
notes more recently that the FCC (Federal Communications Commission) has now legalised UWB, but a “single cohesive standard” is still under development.
The 802.15.3a group is working on a single approach to UWB specification known as “multi-banding”. This could become a competitor for Bluetooth – the same low power consumption but much higher data transfer ([15]).
ZigBee: ZigBee has similar characteristics to Bluetooth, its main design goals are low power, low cost and robustness. (Legg 2004 [17]). Power is saved by going into “sleep” mode with the capacity to switch into active mode in less than 15msecs. Bluetooth has a more significant switching mode delay of around 3 seconds. Its range can be extended to an impressive 134m operating on a low data transfer rate. ZigBee operates at same frequency band as Bluetooth using frequency-hopping spread spectrum (FHSS) but fewer hops (25 every 4MHz), and slower data transfer at 250Kbps. Kardach [18] says chips are being
manufactured by major semiconductor companies and some are being produced for “select” customers.
Bluetooth: Goals and Vision
Haartsen et al [7] provides an overview of the beginnings of the standardization of Bluetooth. At the time there was a lack of “universal framework that offers a way to access info based on diverse set of devices.” The Bluetooth SIG was formed in 1998 to realize this goal. By bringing different technology sectors together, they hoped to provide “a universal and ubiquitous connectivity solution between computing, communication and supporting devices.” ([7]).
The modern version of the Bluetooth SIG ([9]) is slightly different.
Instead of being run on a volunteer basis by employees of member companies, there are now working groups focussing on specific areas – engineering, qualification and marketing. Their mission statement: “Develop, publish, and promote the preferred short-range wireless specification for connecting mobile products, and to administer a qualification program that fosters interoperability for a positive user experience.”
Key characteristic of Bluetooth is to combine “usability models based on functions provided by different devices.” ([7]). A solution could be reached by providing a wireless connectivity, networking and application framework. It was specified from the outset that the Bluetooth module should also be small enough for integration into portable devices. This should not “significantly compromise battery lifetime of the device.” The specifications also recognised the dynamic nature of Bluetooth networks for detection and connection establishment. ([7]).
Despite these promising ideas Bluetooth has taken some time to
overcome teething problems. The Bluetooth Specification Version 1.2 released at the end of 2003 detailed faster data connections and attempts to reduce
interference with other wireless devices (Lemon el al 2003 [26]). This could still be an issue due to the overcrowding of the specified frequency band ([19]). There has also been a slow adoption of Bluetooth over the years partly due to its
complex usage “profiles” (discussed under Profiles). Costs have also been slow to come down as well as the data rate still limited to 500Kbits/s. It is suggested however that developers should look to maximise Bluetooth capabilities rather than immediately look to another technology ([23]).
Bluetooth the specification is divided into two main sections: how the technology works (architecture) and how the technology is used (profiles). (Jones et al 2002 [20]).
Bluetooth Specification: Architecture
Radio and protocol definitions define an application framework that is
interoperable with other protocols and protocol stacks. Compliance specifications apply to vendors and developers who wish to display the Bluetooth Logo ([7]).
A Bluetooth unit contains RF radio, baseband controller and
microprocessor on single CMOS integrated circuit ([20]). The single chip design was required to meet the low cost, power and size requirements. Transmission occurs on the unlicensed Industrial, Scientific and Medical (ISM) band using Frequency-Hopping Spread Spectrum (FHSS) [7] with retransmission always on different channel [20].
Bluetooth Network Topologies
Bluetooth supports “point-to-point and point-to-multipoint connections” ([7]).
Enabled devices have automatic communication – no specification needed. When units come into range, a connection will be established and detection if there is any information to be passed. Pairing is the initial communication process when creating a new connection with an unknown device ([19]). Once connections
have been established, virtual channels are defined using pseudo-random hop sequences.
Piconet and Scatternets: A piconet consists of two or more radio devices sharing the same channel. There is a master device and one to seven slaves.
Inactive slaves will exist in stand-by mode (this is to save power – [7]). Masters initiate Bluetooth communication links – but slave may request to become a master. Time Division Multiplexing is used to divide up time slots. Multiple overlapping piconets are called scatternets ([20]). Too many piconets result in more collisions, and can lead to falling data rates. Algorithms for forming scatternets and symmetric procedures for establishing connections are both popular topics of Bluetooth research (Bhagwat 2001 [21]).
Bluetooth Specification: Profiles
Hopkins 2003 [24] describes the Bluetooth stack is a “controlling agent” that implements Bluetooth protocol and allows one to control a device from a software point of view. This stack comprises of layers (sub-protocols) and profiles, allowing the user to communicate with other devices and control their own device (Figure 1)
Figure 1: Outline of the Bluetooth Stack (modified from Boling 2002 [27])
The Host Controller Interface (HCI) interfaces between the host and controller.
LMP (Link Manager Protocol) is the protocol that handles link establishment between Bluetooth devices and BB (Baseband) enables the physical radio frequency (RF) link between Bluetooth units. (Hopkins 2003 [24] and MSDN 2004 [28]).
The Logical Link Control and Adaptation Layer (L2CAP) handles data
transmission from higher layers and SDP (Service Discovery Protocol) discovers Bluetooth devices in the surrounding area. RFCOMM (Serial Cable Emulation Protocol) is the layer providing for the creation of virtual serial ports and stream data.
Application Developers can utilize underlying capabilities of Bluetooth Stack via an API – Application Programming Interface (Gratton 2003 [22]).
OBEX is the object exchange protocol supporting upper-level applications; for example data synchronization, file, object or data transfer ([23]).
Bluetooth Profiles define which features of the Bluetooth Stack are required and how they will be used ([22]). These form basis for Bluetooth interoperability. A profile can be visualised as “a vertical slice through the
protocol stack.” and may depend on other profiles. ([20]). Some examples of profiles are: Generic Access, service discovery, serial port, headset, generic object exchange, and object push profile. ([21])
Senese 2001 [23] mentions that profiles have been carefully defined in the Bluetooth specification and are applications that allow short-range
communication.
A control processor or host facilitates upper layers of protocol and application software. Thus despite the size of the microcontroller (ARM is often used, could be merely a 64kB processor) both application and stack must fit on this hardware. [21]
Data rate of devices is managed by hardware and cannot be directly specified by application. ([23]). Beutel et al. 2001 [25] implemented a Minimal Based
Bluetooth-Based Computing and Communication Platform - small but generic wireless networking node.
Bluetooth Application Development
Before developing end-user applications, application developers will need to perform the following steps according to Senese [23]:
1. The Protocol stack and operating system must be integrated.
2. Integrate transport driver that interfaces the stack to the radio hardware ( UART or USB hardware interface are suggested).
3. Support software developed that is responsible for establishing connections to other Bluetooth devices.
The Bluetooth Stack can now be used to implement the application, system control and bidirection data flow.([23]).
The Bluetooth SIG was promoting “development of applications that work over either IR or RF. The application developer should not care what medium is used.” ([7]). Unfortunately this is not necessarily the case. According to (Hall 2003 [34]) there is no definition in the Bluetooth Specification
suggesting how application on a higher level should “talk” to the underlying
stack. The Bluetooth Specification makes no mention of how a simple exposed API is defined leading many manufacturers to add their own unique APIs.
There are an abundance of vendors who provide Bluetooth Stacks and APIs in the form of Standard Development Kits (SDKs).
Among the more widely used:
• Widcomm: one of the few that allows access and modification directly with its Bluetooth Stack,.
• Socket Communications: allows developers to create Windows CE applications,.
• Sourceforge 2004 [29] also provides a long list of Open Source Bluetooth APIs to use on a variety of platforms in a number of different environments and
programming languages. Some perform more specific tasks than others – BTNode System Software for example is an autonomous wireless
communication and computing platform based on a Bluetooth radio and a microcontroller used for research in mobile and ad-hoc networking.
• Nokia, Motorolla, Phillips, Intel, IBM – all major manufacturers provide their own unique APIs.
Listing provided by Palo Wireless 2004 [30].
A challenge facing application developers is that there is no standard Bluetooth API. [24]. The Java API is one of the few standard industry APIs. The authors claim that for any C/C++ based Bluetooth SDK, the developer is “at the mercy” of the vendors.
Cole 2004 [31], suggests that what is needed is a common applications programming interface that “is industry wide, crossing boundaries of OS,
application and most importantly, CPU architecture”. He claims, that despite the reduced number of competing groups backing specific architectures; OS and development tools, developers are still in a difficult position. Each different product requires “rethinking all the way down to the programming level” - having to consider which platform is being used.
Evers 2004 [32] reports that Koninklijke Philips Electronics and Samsung Electronics are launching a software standard that all manufacturers are invited to employ. They have noted there is no standard API for developers to write to. The Universal Home Application Programming Interface (UH-API) has been created
to attempt to prevent developers having to write “the unique hardware requirements of every product being developed.”
The Embedded Linux Consortium 2004 [33] also attempting to create a standard Embedded-Linux API. They envision a platform standard for API's that will allow interoperability and successfully challenge “proprietary or home-made solutions.”
Working with Windows APIs isn’t “easy, clean, or quick.” ([27]), due to the flexibility of the Bluetooth standard and the complexity of the service discovery protocol (how device’s available services are communicated).
Windows Bluetooth Stack
The two interfaces for interacting with the Windows Stack are Winsock API and virtual serial ports: Winsock is the recommended method (according to [27]).
This is because despite their familiarity to programmers, serial ports have issues with driver names and their implementation is often unique. They are, however, useful when legacy support is needed.
Winsock
Winsock defines a network programming interface for Windows, based on the
“socket paradigm” popularized in USB Unix. (Quinn 1998 [35]). Winsock 2.0 is the most current version providing more functionality than the previous version, Winsock 1.1. Sometimes Winsock 1.1 is preferred because of its smaller size.
Windows CE utilizes Winsock 2.
WinSock 2 has two interfaces:
API (application programming interface): shielding developers from the underlying layers. SPI (service provider interface): allowing transparent extensions to the Winsock stack.
There are also two types of Sockets: Stream (Connection Orientated Connection) and Datagram. A Stream Socket can be described as a “data pipe” – once initial connections have been set up, data can be sent back and forth without the need for further addressing. Datagram Sockets are compared to “mail slots” where separate packets of data are sent to specific addresses. ([27]).
Using Winsock for Communication over Bluetooth:
An example of the steps needed to connect two devices is given in [27]. Stream socket connections are described in terms of the Client-Server model. Initially both client and server must each create a socket for connections.
Server side:
• Socket is bound to an address. The address structure includes device’s 48 bit Bluetooth address and most importantly the Bluetooth RFCOMM Channel. This is published in an SDP Record for the other devices to look up – devices will need to know which port to connect to when attempting to connect to the server.
• Server socket is placed in listen mode, now accepting incoming requests. It will block until client socket attempts to connect to server address.
• When a connection is accepted, a handle to new socket (connected to client) is returned allowing communication with client. (the method contains buffer that receives client address)
Client side:
• First device discovery must be performed. List of devices and their services should be built up before communication can be attempted.
• Socket should connect to a selected server – just connects, doesn’t need to bind and accept. The client must know server RFCOMM channel or GUID (specific unique identifier for a particular service offered on a particular device).
• Sending and receiving data – both client and server will have sockets by now to use for communication.
• Finally once communication has terminated both sockets must be closed..
There are several design considerations when programming with Winsock 2.
Asynchronous functions are not supported but a socket can be placed into non- blocking mode. Windows CE does support multithread programming.
String fields are char fields in socket structures, not Unicode, one may need to convert. Windows CE does not provide a method to expose a raw socket – and thus cannot deal directly with IP layer of TCP/IP protocol, making pinging impossible. To send an echo request, ICMP Internet Control Message Protocol must be used. ([27]).
Conclusion
Bluetooth Technology could provide a possible solution to monitoring animal interactions automatically. A number of techniques already exist but none can track interaction and only some are automated. Tagging requires unnatural human interaction. Other networking technologies exist but Bluetooth is one of the most currently viable. Bluetooth Application Development is an issue due to a lack of API standardisation.
Current Animal Tracking Methods
[1] WWF’s Polar Bear Tracker. “Tagging Polar Bears”. 3 May 2004. World Wildlife Fund for Nature. 18 May 2004.
http://www.panda.org/about_wwf/where_we_work/arctic/polar_bear/tagging_bears.cf m
[2] Stephenson J.P. “News From Beyond the African Elephant Programme.” Elephant Update – Recent News from the WWF African Programme June 2003. Published by WWF- World Wide Fund for Nature, Switzerland. 18 May 2004.
[3] Amstrup S.,York G., McDonald T., Nielson R., Simac K. “Detecting Denning Polar Bears with Forward-Looking Infrared (FLIR) Imagery” . BioScience Vol.54 No.
4. p 337-343. April 2004.
[4] Alibhai S., Jewell Z. “Hot under the collar: the failure of radio-collars on black rhinoceros Diceros bicornis”. Oryx Vol. 35 No. 4 p284-288. October 2001.
[5] Steiner I., Burgi C., Werffeli S., Dell’Omo G., Valenti P., Troster G., Wolfer D., Lipp H. “A GPS logger and software for analysis of homing pigeons and small mammals.” Physiology & Behaviour. Vol. 71. Issue 5. p 589-596. December 2000.
[6] TVP Positioning AB “Televilt GPS collars – GPS tracking collars, animal tracking transmitters and telemetry datalog.” April 2004. 2 May 2004.
<http://www.televilt.se/code/gpsmain.asp>
Introduction to Bluetooth
[7] Haartsen, J. Allen, W. Inouye, J. Joeressen, O. Naghshineh, M. “Bluetooth:
Vision, Goals, and Architecture.” Mobile Computing and Communications Review.
Vol. 1, No. 2. 1998.
[8] Franklin C. “How Bluetooth Works”. How Stuff Works 2004. 7 May 2004.
<http://www.howstuffworks.com>
[9] The Official Bluetooth Website May 2004.: Bluetooth FAQ. Bluetooth SIG.
<http://www.bluetooth.com/help/faq.asp>
Other Wireless Technologies
[10] Barnes S. “Under the skin: short range embedded wireless technology”.
International Journal of Information Management.p 165-179. No 22 2003.
[11] Dursch, A. Yen, D. Shih, D. “Bluetooth Technology: an exploratory study of the analysis and implementation frameworks.” Computer Standards & Interfaces 2004.
Available from <www.ElsevierComputerScience.com>
[12] Dividney, G. “A Summary of Short-Range Wireless in 2003” TechOnline 2004
<http://www.techonline.com/community/tech_topic/bluetooth/25530>
[13] Suvak D.“IrDA and Bluetooth: A Complementary Comparison.” Extended Systems Whitepaper 2000.
[14] Pico Communications Whitepaper: “Cats and Dogs living together : A Comparison of Bluetooth and Wi-Fi.” October 2001. <http://www.pico.net>
[15] Official Ultrawide Band Website 2004. <http://www.uwb.org/faqs.html> 2 May 2004.
[16] Bowles M. “Ultra-WideBand: It’s Not Just Hype.” TechOnline
<http://www.techonline.com/community/home/23148> 2004. 19 May 2004.
[17] Legg G. “ZigBee: Wireless Technology for Low Power Sensor Networks.”
TechOnline <http://www.techonline.com/community/home/36561> 2004. 19 May 2004.
Bluetooth in Particular
[18] Kardach, J. “Bluetooth Architecture Overview” Intel Technology Journal 2nd Quarter 2000. Chao, L. Ed.
http://www.intel.com/technology/itj/q22000/articles/art_1.htm
[19] Dursch, A. Yen, D. Shih, D. “Bluetooth Technology: an exploratory study of the analysis and implementation frameworks.” Computer Standards & Interfaces Vol?
2004. Available from www.ElsevierComputerScience.com
[20] Jones S., Badgett N., Eskala N., Paul J., Seuk Yun S. “Emerging Wireless Technologies – Bluetooth and Ultra-Wideband.” Institute of Wireless Innovation Whitepaper November 2002
[21] Bhagwat, “Bluetooth: Technology for Short-Range Wireless Apps”. IEEE Internet Computing May/June 2001 p96-103.
[22] Gratton, D. “Bluetooth Profiles: The Definitive Guide.” 2003 Pearson Education, Inc. Publishing as Prentice Hall Professional Technical Reference, New Jersey.
Application Development
[23] Senese, B. “Bluetooth: Application Development Made Easier”. Wireless Systems Design, Vol. 6 Issue 4. April 2001.
[24] Hopkins, B. Anthony, R. “Bluetooth for Java.” Apress. New York. 2003 [25] Beutel J., Kasten O. “A Minimal Bluetooth-Based Computing and Communication Platform”. Technical Report May 2001.
<http://www2.inf.ethz.ch/~kasten/research/papers/iswc2001.pdf>
Retrieved 2 May 2004.
[26] Lemon S., Lawson S. “Whatever Happened to Bluetooth?”.
IDG News Service. 14 November 2003. 20 May 2004.
<http://www.pcworld.com/news/article/0,aid,113404,pg,1,00.asp>
[27] Boling D. Programming Microsoft Windows CE .NET , Microsoft Press.
Washington. 2002 (third edition).
[28] MSDN Library “Bluetooth Stack Architecture”. Microsoft Corporation. 8 April 2004. 30 March 2004. <http://msdn.microsoft.com/library/default.asp?url=/library/en- us/wcebluet/html/ceconbluetoothstackarchitecture.asp>
[29] SourceForge.Net “Open Bluetooth Project.” 2004. Open Source Development Network 20 May 2004. <http://sourceforge.net/>
[30] PaloWireless Bluetooth Resource Center. “Bluetooth Development Tools”. Palo Wireless. 19 May 2004. 20 May 2004.
<http://www.palowireless.com/bluetooth/devtools.asp>
[31] Cole, B. “In search of a common API for connected devices”. Embedded.com 5 Feb 2004 (16:00 PM) 20 May 2004.
<http://www.embedded.com/showArticle.jhtml?articleID=17602108 >
[32] Evers, J. “Tech Giants Team for Consumer Devices”. IDG News Service 6 Jan 2004. 19 May 2004.
<http://www.pcworld.com/news/article/0,aid,114109,00.asp>
[33] Embedded Linux Consortium. “Questions and Answers”. 2004. 12 May 2004.
<http://www.embedded-linux.org/qa.php3#Q2>
[34] Hall M. “Windows Embedded and Wireless Technologies in Windows CE.NET.”
Embedded Systems Crash Course – video presentation. Microsoft Corporation 2003.
[35] Quinn B. “Windows Sockets Network Programming” Sockets.com 3 Nov 1998.
12 March 2004. <http://www.sockets.com/>
[36] Song, C. Wu, S. ‘BlueTrak’ Presentation and Whitepaper. University of Virginia.
2002. Obtained from Gorman, M. 2001-2004. Invention and Design. 7 July 2004.
<http://repro-nt.tcc.virginia.edu/classes/tcc315_2003/materials/bluetrackNCIIA.pdf>
