From Sensors to Location Authority: a Design Notation
C.W. Johnson
T he Australian National University
Canberra, A.C.T. 0200, Australia
Chris.Johnson@anu.edu.au
11
June 2004: text of a poster
paper submitted to UbiCOmp 2004
A specialised graphical notation helps designers to express designs of distributed location-awareness systems that use sensors, events, accumulated abstracted evidence, evidence resolution, and location authority databases. The notation is a specialisation of data flow diagrams.
design notation, location-awareness, sensors, location authority, evidence resolution
A commonly met structure for location-aware mobile computing applications is the combination of wireless devices acting as identity badges, sensors or beacons, with data reduction or analysis processes that rely on a location authority to relate device identity directly to a description of its location, or indirectly through relations to other devices.
Although implementers of such systems share a small number of common information models (abstractions of location and mobile entities) and classes of hardware and software elements, the basic elements and the choices of system architectures in implementations of such systems vary widely. For example, sensors range from very short range anonymous proximity, to GPS coordinates; location descriptions range from coordinates to unrelated unique names. Systems differ in the direction of the sensors (in walls, or in mobile devices); in the type and form of description of a “location”; and in the network distribution of their logical processing and storage components.
But the fact that we can share descriptions of choices and architectures shows we understand a common set of ways of designing systems. For example: one common structure is to filter, aggregate and transform data from sensors; another is to maintain a data relationship between locations and other entities (a location authority). A third is the accretion and resolution of evidence, where relatively high frequency streams of event data from dumb sensors are refined, filtered, and aggregated into lower frequency streams of more meaningful evidence; less frequently, chosen accumulations of evidence are resolved to a conclusion.
These four concepts of (1) a layer of dumb and smart sensors, (2) transformation of sensor data to accumulated evidence, (3) resolution of evidence to conclusions, and (4) a Location Authority, can be used to express the design of systems in a simple way for system implementers to understand, share and compare their ideas.
Design notations aid human thinking. A design can expose shortcomings and potential faults in systems, before expensive programming and debugging. A notation can also express a toolkit of ideas (design patterns). This is true even when there is no automated tool in support.
To design a location awareness system we need to show its component processing, storage, and flow elements, their types, and the scope of their instances. The elements are:
The functions of a Location Authority are due to Shafer[5]. Layers of function within the sensor layer are seen in Context Widgets[4] and the Merino architecture[2]. This notation can be applied with other processing models, but the evidence/resolution model[1] is common, often being used without being described as such (e.g. Narrator[7]). Data Flow Diagrams are a textbook design notation. More specialised forms are found in automated tools such as Box[6], without the specific elements that we found useful.
Figure 1 shows the major components of an application to display a map on a mobile device showing the location of visible Bluetooth wireless beacons, in a large building[3].
The roles of key graphical components (stored relation, resolver etc) have been added in the margins.
Dashed enclosures are added in two layers: for logical modules (application, location authority) and components’ network distribution (not shown here). These views of the designed architecture expose important aspects to scrutiny: performance (amounts of data over networks), and privacy: whereabouts sensitive data is stored or is sent on networks.
I acknowledge the support of
the Smart
Internet Technology Cooperative Research Centre, and colleagues in SIT Research
group,
1. Kay, J, Kummerfeld, B, Lauder, P. Managing private user models and shared personas, UM03 Workshop on User Modeling for Ubiquitous Computing, p 1-11
2. Kummerfeld, B, Quigley, A, Johnson, C.W., Hexel, R. Merino: Towards an intelligent environment architecture for multi-granularity context description, Workshop on User Modeling for Ubiquitous Computing 2003.
3.
Quigley, A, Ward,
B, Cutting, D, Ottrey C, Kummerfeld, B. BlueStar,
a privacy centric location aware system,
IEEE PLANS 2004, Monterey CA, USA
4. Salber, D, Dey, A.K., Orr, R.J, Abowd, G.D. The Context Toolkit: Aiding the Development of Context-Enabled Applications, Proc 1999 Conference on Human Factors in Computing Systems, p 434-441
5. Shafer, S.A.N. Location Authorities for Ubiquitous Computing, Proc. 2003 Workshop on Location-Aware Computing, p 13-15
6. Vina, V. Box. Open System to Design your own Network, UbiCOMP Adjunct Proceedings, 2003, p 127-130
7. Wilson, D and Atkeson, C. The Narrator: A Daily Activity Summarizer Using Simple Sensors in an Instrumented Environment, UbiCOMP Adjunct Proceedings, 2003, p 141-144