Friday, June 28, 2013

Internet of Things and Distributed Cyber-Physical Systems

I was delighted to see that the latest round of Presidential Innovation Fellows announced recently by the White House includes not only our very own Dr. Bev Woolf, of the Computer Science Department at UMass Amherst, but also two Fellows in Disaster Response & Recovery and also two in Cyber-Physical Systems in the list of 43 new Fellows.

Coincidentally, a paper written by Professors Tilman Wolf, Michael Zink, and me, and entitled: "The Cyber-Physical Marketplace: A Framework for Large-Scale Horizontal Integration in Distributed
Cyber-Physical Systems," will be presented on July 8, 2013, in the opening session, following the keynote, at The Third International Workshop on Cyber-Physical Networking Systems, in Philadelphia.  In our paper, we acknowledge support from the National Science Foundation on our project: Network Innovation Through Choice.

As we state in our paper: Cyber-Physical Systems (CPS) promise to provide technical solutions to solve some of the most important societal and environmental problems. By combining sensing and response actions in the physical domain with processing and communication functionality in the computational domain, CPS can provide novel solutions in the areas of health care, transportation, energy, disaster response, manufacturing, defense, etc. It can be expected that cyber-physical systems will become ubiquitously deployed in the coming decade as users demand more “smart” solutions in their environment.

Traditionally, cyber-physical systems have been considered to be tightly integrated solutions that are aimed at specific application domains. Typical examples include industrial control, flight control in airplanes, agriculture, etc. These traditional application domains of CPS are often characterized by safety concerns and tight performance constraints. In parallel, a broader view of cyber-physical systems has emerged, often termed distributed cyber-physical systems (DCPS) orthe “Internet of  Things,” where interacting components are more loosely coupled and real-time constraints and safety concerns are less dominating. Application domains from this space include home automation (e.g., lighting and temperature control), smart appliances (e.g., operation based on energy availability and price), smart traffic control, etc.

In our paper,  we argue that in order to develop a common infrastructure for a large-scale DCPS, it is critical to develop a framework that allows for horizontal integration of CPS components. We discuss the vision, requirements, and design of such a framework. We present several specific technical challenges that need to be addressed in this context as well as potential solution approaches. We also provide examples to illustrate how our framework can lead to innovative new uses, from weather sensing to smart transportation,  that can have meaningful impact on practical problems.