Documentation : Hardware | Software | Tutorials    Wardrobe : Jackets | Applications
Papers : Notes | Proposals | Research    Administration : Stock | Minutes | Progress

CITY Workshop - Glasgow, 11th-15th March 2002

"Into the streets"

Introduction

At this workshop we carried out fundamental research to explore the options to enable the CITY project to move from inside the Lighthouse into the streets of Glasgow. This page describes research into GPS coverage and wireless data communications. Much modelling work was also carried out and more information on this can be found here.The opportunity was also taken to progress the ultrasonic positioning system in the Mack Room.

GPS Testing

Reference Positions.

Readings were taken at three positions on the Lighthouse roof (see Fig 1). These positions are intended to ensure that the GPS WGS84 reference system is correctly aligned with whatever models are used for the project.  A poor dGPS signal was only available at two of these positions as the west of Scotland is beyond the normal range of the existing dGPS radiobeacons  - see image.

.

mc_roof.jpg (47082 bytes)

Fig 1 - GPS survey positions

The resulting positions along with variation/accuracy figures are shown below. Variation readings for a typical position in Bristol are shown for comparison. 

Period

Latitude 55deg North Longitude 4deg West

50% variation (metres)

95% variation (metres)

Comments

Lighthouse 1

24hr

51.56931min 15.32760min

3.93

13.70

Lighthouse 2

12hr

51.58651min 15.32647min

3.12

10.43

no dGPS

Lighthouse 3

24hr

51.58813min 15.31612min

6.32

16.11

mostly dGPS

Bristol

24hr

1.40

4.41

Table 1 - GPS Survey results

The Lighthouse variation figures were disappointingly large - however accurate positions were obtained for surveying purposes. It is hoped that these figures - which can also be interpreted as measuring accuracy - will improve substantially when the new dGPS transmitter opens near Stirling ('early' in 2002).

GPS Coverage

GPS coverage tests were undertaken in the streets around the Lighthouse. Severe problems were experienced due to the urban canyon effect. While fixes could be obtained at street intersections, these were typically using only three satellites and hence the accuracy was between 25m and 35m. Similar fixes could also be obtained in front of low buildings such as Princes Square and Borders. Fortuitously five satellites lined up along the east-west axis during one test enabling a four satellite fix to be maintained along Argyle Street - this was exceptional. It is recommended that in the short term a more GPS friendly location should be found for workshops/user tests - perhaps the House for an Art Lover. Later - see GPS Satellite Coverage

Communications

Two wireless technologies were tested - 802.11 and GPRS.

Fig.2 shows, in green, the external coverage from an antenna placed in a north facing third floor window in the tower next to the Mack Room. In addition adequate coverage was obtained throughout the Mack Room. No coverage was obtained inside the other buildings off Mitchell Street.

No changes to the existing software are required to extend the 802.11 coverage outside.

802_coverage.jpg (51864 bytes)

Fig. 2  802.11 coverage (external)

A Bluetooth link was used to pair Vee's Jornada with a GPRS enabled Ericsson T39 mobile 'phone. The 'phone communicated via orange.net and a BT landline to the 802.11 access point in the Mack Room. A hard wire conection to another 802.11 access point enabled connection via 802.11 to a server running on a laptop. To break through the firewall(s) in this chain, all communication was via Port 80. Should we wish to continue using GPRS a robust http protocol would need to be implemented. Extensive testing was impractical in the time available, however data communication was established and the viability of this chain proven. Interestingly Bluetooth and 802.11 performed satisfactorily within feet of each other. Subsequently we discovered that we had been using the Orange WAP service and after a somewhat convoluted setup process connection was established using Orange Internet. At the time of writing investigation continues into the most appropriate protocols to use.

Mack Room Positioning System

The opportunity was taken during this workshop to progress the development of the positioning system in the Mack Room by:-

Disappointingly no significant improvent in the 0.5m 50%; 2.5m 95% accuracies were obtained from the software changes, or by recabling. Indeed the increased ultrasonic levels in the room may have led to an increase in spurious reverberant signal detection. We conclude that the system is probably at it's limit of development.

The beacon tests are summarised in the table below. The graphics show the area of coverage at waist height (in colour) with the square box representing the test cubicle:-

Infra Red R.F. Ultrasonic
Single beacon placed under top of cubicle with two adjacent transmitting LEDs.

Handheld receiver.

R.F. Pinger, with antenna removed, placed on top of cubicle.

Standard handheld receiver (with helical coil antenna).

Transmitting transducer placed under top of cubicle.

Standard handheld receiver.

IR_beacon.gif (1377 bytes)

RF_beacon.gif (1420 bytes)

US_beacon.gif (2085 bytes)

Required explicit action to ensure IR beam was intercepted. Contention with positioning system 'ping' requires redesign of ultrasonic receiver with consequential rise in power consumption. Simple solution with control over zone size (figures in graphic represent the approximate distance from the transducer in metres).

Requires  additional ultrasonic chirps and hence marginally lengthens transmission time.

It is recommended that:-

  1. Ultrasonic beacons are used to enhance the current installation with separate zones for each of the cubicles, and the two ends of the room;
  2. The 2D positioning to the right of the screen is abandoned and replaced with a 1D system which will place the user at a position along a predetermined path;
  3. Positioning to the left of the screen should be retained and optimised.

n.b. The R.F. Pinger beacons are also able to be set up with ranges of 10m and 100m as well as 2m. These can be used where ultrasonics or GPS are not available.

The resulting system is illustrated below:-

mcpos_loc.gif (26502 bytes)

We have the option of reporting a unique nominal position for each zone when entered by Vee. Alternatively a location can be reported. A hierarchical structure for position reporting suggests that a position should be sent. It is intended to implement this arrangement before user tests commence.

 


This page last updated April 26, 2002
The material displayed is provided 'as is'. Contents may not be reused without prior permission.
For problems or questions regarding this web contact Cliff Randell.