Location and Position

Location and position have been widely investigated for their use in context-aware systems, as evident from many projects. When discussing location and position, issues such as co-location and proximity are also relevant.

For sensing location outdoors GPS and dGPS are most popular and easy to use [Hofmann,97], [Letham,01]. However GPS has a long boot-up time (typically 30 seconds to one minute). This time can be reduced by using further information gained earlier or over a data connection. The output of a GPS location system provides the position of the device. Depending on the number of satellites which are visible the

accuracy is within a few meters and with dGPS in the region of centimetres. Most devices support the standardised NMEA0183-format that is exchanged over a serial line protocol. Most GPS systems offer additionally a propriety protocol that is more powerful.

Another option is to use information provided by a cellular network, such as the information about GSM base stations in range and their link quality or just the GSM- cell booked in. The location information is usually only supplied to services in the network [Nakanishi,00], [Park,02]. On some GSM modules or phones the information about the base station booked in and the signal strength can be read out and captured over a serial line protocol (e.g. debug mode of an Ericsson PF768 phone).

In certain setups radio beacons can also be used to provide location information. Here both specific hardware and software is used or the location system is built on top of an exiting infrastructure (e.g. as in GUIDE using WaveLAN [Cheverst,00a]). Further issues on outdoor location systems are discussed in [Bulusu,00].

Another method is to look at what TV and radio stations can be received (e.g. using Radio Data System – RDS), on which frequencies and with what signal strength. This can be realised using a radio and/or TV module where this information is accessible. The location is then determined by matching the reception patterns with reception patterns of which the location is known. There are no commercial products available using this technique but it seems to be an interesting option for systems that include a radio or TV receiver, such as car radios or TV sets.

Comparing these methods GPS offers the service to find the location anonymously, whereas in the other systems it is very much a design decision whether or not a central system is involved in finding the position. The accuracy of the position gained from outdoor location systems can be ranging from town level down to centimetre level. In indoor scenarios where coverage of a building is required different technologies are available and deployable, in particular IR-beacon systems [Butz,00], RFId-Tags, and ultrasonic location systems [Ward,97], [Nissanka,00]. Further approaches are researched, such as WaveLan triangulation, location systems based on existing infrastructure, and RF-beacons [Bahl,00], [Small,00]. When setting up such systems

the design decision can be made as to whether or not anonymous location finding is supported. This usually comes down to the question whether the location is calculated on the local device or within the network. The accuracy of indoor systems is very much dependent on the environment [Regenstein,01].

In our experiments and from reports of other researchers it appears that most indoor location approaches provide very high accuracy under laboratory conditions, but still perform poorly in real environments where people walk around, doors are opened and closed, laptops and PDAs use WaveLAN and Bluetooth and where CRT-screens are switch on and off. Exceptions are the AT&T active bat system [Wart,97] and the MIT cricket location system [Nissanka,00] which are not commercially available.

In order to sense co-location the approach of using radio beacons that can adjust their outgoing signal, or measure the strength of the incoming signal, can be taken. By making the communication range adjustable the degree of co-location can be selected. As shown in the Smart-Its project [SMART,02] this can give an accuracy of about a metre most of the time, however changes in the environment introduce significant errors. Co-location between devices and the environment can also be realised using RFId technology with long range readers. The use of strong readers and large antennas is very disputable in an environment where people live or work.

Location sensing is somehow different from the sensors discussed earlier. The aim of most location systems is to offer a ‘sensor’ that gives meaningful symbolic or geometric location information. This information is then most often used as a trigger, or an index to access further information.

3.5.5 Magnetic Field and Orientation

Different types of sensors are available to detect magnetic fields. Some are designed to detect the earth’s magnetic field whereas others are constructed to detect the proximity or change of a generated magnetic field. Hall-sensors detect the flux of the magnetic field applied.

Sensors that detect the earth’s magnetic field are the basic building blocks for an electronic compass. The output is related to the direction of the magnetic filed and can be used to figure out which direction is north. These sensors are also available

combined with electronic circuits in one component that provide this information on a higher level [Honeywell,02], [Philips,02]. Advanced modules offer information similar to a compass, so the direction of a device or of a movement can be determined.

In our experiments we realised that in modern environments (e.g. offices with computers monitors) this sensor can give false information. Nevertheless there are many application areas where the orientation is of significant value.

3.5.6 Proximity, Touch and User Interaction

Similar to the argument introduced with movement and acceleration it is also a basic observation of the real world that people often interact with things by touching them. Sensing that a user interacts with an artefact can be implemented in various ways and also with different levels of detail, ranging from the mere fact that the user touched the device, to the way the artefact is hold.

Before the user can touch an artefact they have to come close to it; therefore sensing the proximity can offer information that user interaction is likely to be ahead. A very simple way to sense proximity is to use a capacitive sensor. Such capacitive sensors are available off-the-shelf and offer a digital signal when approached and a threshold is crossed. For larger settings or to integrate them into artefacts, such capacitive sensors can also be built into objects using metal sheets and a driving electronic circuit. Proximity sensors that offer the distance of an object or the user hand are also available based on light. The analog output is related to the actual distance of an object in front of the sensor.

Humidity sensors can also be used to provide information on the proximity of users. The humidity rises when users approach an artefact. This rise however is extremely small, and high quality sensors are required to measure that change.

Conductive surfaces on artefacts can be used to get information on touch [Hinckley,99]. These surfaces can also be used to measure the skin conductance of the user and to some extent muscle tension. There are additional amplifiers required to provide the signal in a way that it can be read by a microcontroller. The values on skin

conductance and muscle tension are also dependent on how strong the users grip on the artefact is, because the measurement also includes the transition resistance.

A further option to measure user interaction and touch is force sensitive resistors. These surfaces change their resistance according to the force applied to them. Putting them on the surface of an object, means that the object can be use to measure how strong the grip of the user is. Using strain gauges on the structures of artefacts can also provide information about the way a device is held. By the user’s grip the artefact is minimally deformed and that can be measured, as our experiments with a ball pen showed.

Other sensors such as light sensors, temperature sensors, and CMOS-cameras can also be used as the source for information about touch and proximity. Typically light sensors are covered when a user holds the device and when held for longer the device also heats up from the body heat.

Acquiring information about touch and proximity can be utilised to realise when the device is being used or in the case of proximity to anticipate that the device will be operated in the near future. By these means these sensors can reduce energy consumption significantly, especially for devices that only need to be operational in the user’s hand. Touch and the way artefacts are gripped can also offer further information about the interaction process and in some cases about the user’s emotional state.