A Standard for Real-Time Smart Transducer Interface Wilfried Elmenreich, Wolfgang Haidinger, Hermann Kopetz Thomas Losert, Roman Obermaisser, Michael Paulitsch, Philipp Peti Institut f ¨ ur Technische Informatik Technische Universit¨ at Wien Treitlstrasse 1-3/182-1, Vienna, Austria {wil,wh,hk,tl,ro,mp,php}@vmars.tuwien.ac.at Research Report 35/2002 Abstract In order to handle the inherent complexity of the mul- titude of available different transducer components, a generic interface approach is necessary. Such a univer- sal smart transducer interface should provide precisely defined interfaces between smart transducers and their users, which are simple and precisely specified within the value domain and the temporal domain. The Object Management Group adopted a smart transducer interface standard that incorporates (i) real- time characteristics and functionalities for the smart transducer network (ii) online diagnostic service capa- bility (iii) support for start-up and dynamic configura- tion (iv) a uniform naming and addressing scheme for all relevant data in the smart transducer system (v) a generic interface that enables the smart transducer sys- tem to interact with other systems via CORBA. This paper describes the main concepts and imple- mentation experiences of this smart transducer inter- face. The approach integrates a time-triggered commu- nication protocol with an appropriate access scheme, the so-called interface file system. This interface file system provides a unique naming and addressing scheme enabling access to internal transducer data via a CORBA gateway. 1 Introduction The decreasing cost of computing power enables de- centralization of data processing by providing computa- tional intelligence close to transducers (sensors and ac- tuators). This enables data abstraction necessary for an abstract interface definition. Such an interface definition comprises a major element in software engineering for embedded systems avoiding an emerging crisis in em- bedded systems design due to the increasing number of different transducers in current embedded systems. In embedded system applications with dramatically increasing numbers of transducers, as it is the case in the automotive market, which is marked by an ad-hoc approach of system engineering with a multitude of dif- ferent components (and different interface definitions), a generic approach can significantly reduce the develop- ment costs. Some interface descriptions for embedded compo- nents exist already and are used extensively (e.g., CAN Kingdom [1], OSEK [2], and IEEE 1451.2 [3]), yet do address a precise interface description in the time and space domain and standardization is often supplier spe- cific. System engineering of a CAN-based approach, e.g., can be inherently difficult and lead to unintended system behavior for certain CAN implementation (see e.g. Meschi et al. [4]). In this paper we present a generic approach for the or- ganization of transducers and communication between transducers that provides features in terms of real-time guarantees, complexity management, and maintainabil- ity. Smart transducers integrate an analog or digital sensor or actuator element and a local microcontroller that contains the interface circuitry, a processor, mem- ory, and a network controller within a single unit. A smart transducer transforms the raw sensor signal to a standardized digital representation, checks and cali- brates the signal, and transmits this digital signal via a standardized communication protocol to its users [5]. The possibility of reducing complexity using the con- cept of a smart transducer is of paramount benefit as ba- sic element of a system engineering approach in embed- ded networks, which cannot be valued highly enough, though initially overlooked due to the current supply chain structure. In addition to providing a building block for a suc- cessful management of a smooth software engineering process of embedded systems, a precise interface de- finition facilitates communication system design. We present a communication system with guaranteed time- liness and a deterministic description of communication traffic. Such a description of a distributed transducer system design can alleviate system development. Time partitioning mechanisms on a communication bus de- crease the wiring loom complexity and provide addi- tional benefits such as reduced weight, reduced wiring 1