Towards Middleware Components for Distributed Actuator Coordination Joel W. Branch and Boleslaw Szymanski Department of Computer Science Rensselaer Polytechnic Institute Troy, NY 12180 {brancj, szymansk}@cs.rpi.edu Chatschik Bisdikian, Norman Cohen, John S. Davis, Maria R. Ebling, and Daby M. Sow IBM T. J. Watson Research Center Hawthrone, NY 10532 {bisdik, ncohen, davisjs, ebling, sowdaby}@us.ibm.com Abstract The pervasive presence and availability of sensor and actuator networks creates the potential for widely distrib- uted Internet-scale control systems. Given these systems’ potential magnitude and complexity, new tools will be required to facilitate their composition and operation, especially as they relate to distributed actuator coordina- tion. This paper introduces and advocates our approach to implementing distributed actuator coordination algo- rithms using a middleware framework. 1. Introduction For decades, sensors and actuators have been used in implementing various forms of automated control systems (e.g., climate control or industrial plant control). Recent technology advances (and associated cost reductions), however, have enabled sensors and actuators to become much more connected and pervasive. These trends are further accelerated by factors such as the increasing per- vasiveness of Internet access and new wireless network- ing standards. These factors have enabled sensors and actuators to communicate with each other with growing ease within ever growing networked infrastructures form- ing sensor and actuator networks, or SANETs. Depending on the application, other networked entities such as data- bases, web services, and messaging systems may also be included as sensors or actuators. When combined with large-scale networked systems, the culmination of these advancements define a new generation of highly perva- sive, dynamic, and heterogeneous Internet-scale control systems, which we refer to as SANET control systems. Along with the proliferation of SANET control sys- tems comes an assortment of new research challenges. One significant challenge, and the focus of this paper, is distributed actuator coordination, which involves manag- ing the behavior of actuators that are related via overlap- ping treatment domains. Treatment domains are defined by the spatial extent over which an actuator affects an environment. Focusing on the details of actuator coordi- nation in emerging SANET control systems is important for several reasons. First, overlapping treatment domains create both the potential for actuator interference and op- portunity for cooperative shared task execution. While redundancy among sensors is typically not considered a negative feature, overlapping treatment domains could affect a SANET control system’s ability to satisfy local and global application goals. Second, the relationship between actuators and their corresponding treatment do- mains may be dynamic. This may depend on factors such as the actuator(s) chosen to perform a specific task, in- consistent actuator performance, or the nature of the me- dium to be controlled. To compound these two chal- lenges, SANET control systems will have a magnitude that far exceeds the scale of control systems encountered in prior generations. The number of distributed, transient, and potentially heterogeneous actuators and sensors that will populate SANET control systems will require novel programming abstractions that were previously unneces- sary. Given the motivations above, we propose to address distributed actuator coordination via the Sentire frame- work [1], which supports the composition of extensible middleware for SANET control systems in a high-level, platform-independent manner. In a sense, this paper pre- sents the next step in the evolution of Sentire and intro- duces its extensions for supporting the implementation of algorithms for distributed actuator coordination. We con- centrate our initial research in defining suitable actuator programming abstractions and a service for facilitating coordination among actuators with interfering treatment goals. To the best of our knowledge, there is no previous proposal to address such a coordination scenario using a high-level middleware approach. This paper proceeds as follows. Section 2 describes our research scope and the pertinent research challenges. Section 3 describes the proposed Sentire approach to supporting actuator coordi- nation. Section 4 illustrates the utility of this approach in building an example of a SANET control system. Section 5 concludes with a discussion of related works and future research endeavors. 2. Research scope and challenges Copyright © 2006 IEEE. Reprinted from 3rd IEEE Workshop on Embedded Networked Sensors (EmNets 2006), Cambridge, Massachusetts, May 30-31, 2006, pp. 71-75. This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of IBM's products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to pubs-permissions@ieee.org.