Matthias Lampe Application-level system and tool support for auto-ID 
application development. PhD thesis No. 18787, ETH Zurich, Zurich, 
Switzerland, 2010 


Abstract 

Radio Frequency Identification (RFID) systems have begun to find greater use 
in the consumer object identification market, in industrial automation, and in 
mobile asset and supply chain management, in a wide range of industries such 
as retail, pharmaceutical, and defense. The term automatic identification 
(Auto-ID) applications is used to generalize RFID applications and includes 
other Auto-ID technologies such as barcode or Bluetooth. More complex Auto-ID 
applications also use sensor technologies not only to identify objects but 
also to track their state (e.g. temperature sensors to ensure quality of 
perishable goods). 

Bridging the gap between the physical world of objects (e.g., products and 
logistical units) and the digital world of IT systems to improve existing 
business processes is the main driver of the widespread RFID adoption. 
However, even if the gap between the physical and the digital world is 
bridged, there is still a gap between applications or enterprise resource 
planning (ERP) systems on one side and Auto-ID systems on the other. 
Application developers are faced with several challenges: Instead of 
concentrating on the application or business logic they have to deal with 
Auto- ID specific details such as readers, tags, or specific error sources. 
They also have to duplicate code when developing a new application for 
capturing Auto-ID data from readers, for filtering and aggregating Auto-ID 
data to increase their accuracy, for providing persistent storage and querying 
capabilities, and for reacting to business events such as the arrival of a 
shipment. 

To facilitate the development of Auto-ID applications, this thesis provides 
concepts, programming models, building blocks and tools that abstract from 
Auto-ID specific details and provide the necessary services and the 
appropriate level of reuse. In particular, the contributions are: 

- An Auto-ID Object Model that abstracts from low-level Auto-ID and sensor 
concepts which provides the means for an application to model and represent 
its domain as the base for the application logic. The model focuses on the 
domain of Auto-ID applications, that is, applications whose application logic 
is based on an implicit or explicit model of the physical world and is 
triggered by (near) real time observations of the physical world through 
Auto-ID readers and sensors. 

- A state machine-based programming model that allows defining Auto-ID 
related micro business processes. Typically Auto-ID applications are not 
interested in all the dynamic changes of the observed physical objects. The 
business process definitions allow an Auto-ID application to integrate 
application logic into an Auto-ID infrastructure in order to only report 
exceptionally states related to the objects, their properties and 
relationships. 

- A visual tool-based approach to instantiate, configure and manage an 
Auto-ID infrastructure. Typically an Auto-ID infrastructure consists of many 
different hardware and software components that have to be customized to fit 
the need of a certain application domain. Such an instantiation and 
configuration is a tedious task and requires programming skills in addition to 
Auto-ID and application domain knowledge. The visual tool-based approach 
provides a concrete representation of the application domain and supports 
nonsoftware developers in the different tasks over the lifecycle of an Auto-ID 
infrastructure. 

The evaluation, based on a prototypical implementation of an Auto-ID 
Infrastructure called the Object Monitoring System (OMS) and on several 
representative case-studies, shows that the contributions of the thesis 
provide the right level of abstractions and services to facilitate the 
development of Auto-ID applications in several different application domains. 
The Auto-ID Object Model provides for an adequate representation of the 
application domains. The programming model allows defining a variety of 
Auto-ID related business processes in the application domains supporting fast 
processing of Auto-ID data in a business workflow. The visual tool-based 
approach proves to facilitate the instantiation, configuration and deployment 
of the Auto-ID infrastructures in the application domains especially for 
non-software developers. 

All in all, this thesis provides sound concepts and its contributions help to 
facilitate the development of Auto-ID applications, help to decrease 
development costs and, in general, contribute to the dissemination of Auto- ID 
technologies in industry and other areas of application.