A Semantic Framework for Reconfiguration of Instrumented Cyber Physical Spaces

With recent advances in embedded computing, networking, and related information technologies, it is now feasible to deploy a variety of sensing devices, communication networks and IT services in the real world, creating instrumented smart spaces. Such instrumented physical spaces may contain variety of sensors including optical sensors, acoustic sensors, accelerometers, GPS devices, cell phones, RFID, Motes, as well as specialized sensors such as people-counters and load-cells that enable us to monitor the state of the physical world and its artifacts as well as activities that ensue in such a world. The multimodal data capture devices in these physical spaces are interconnected by a variety of standard (and possibly proprietary) communication networks, including wired Ethernet, cellular, WiFi(802.11b), Bluetooth, Zigbee, WiMAX, MANETS, mesh networks and power-line networks. The ability to integrate sensing and communication platforms with large scale distributed storage/compute facilities and software services enables the creation of Instrumented Cyber Physical Spaces (ICPSs) that are technologically cutting-edge and exemplary of highly-instrumented spaces of the future. A variety of applications and services execute on such ICPS, often simultaneously; examples include asset monitoring, surveillance, social networking, facilities management and emergency response. Services such as medical triaging services, emergency response services, etc. involve human processes and decision making where humans in the loop determine how the application evolves based their awareness of the situation and infrastructure. In these applications, users in different roles execute on varying enduser devices (e.g., desktops, PDAs, cellphones, large scale display) and present information needs that vary in type, quality, and modality requirements. Applications dictate application specific constraints on the timeliness and accuracy/quality at which information must be captured and delivered from the infrastructure. Technologies and software systems deployed currently to manage such instrumented spaces are often designed and engineered for specific applications. Sensor programming frameworks often require knowledge of the low level sensing infrastructure (types and attributes of sensors). Repurposing the infrastructure, its software and hardware resources dynamically to realize different application functionalities is difficult. In this context, we are developing frameworks that can customize the operations of ICPS to meet the varying needs of applications and users. Our approach to developing flexible and efficient cyber physical spaces is based on the principles of computational reflection, including an observe-analyzeadapt philosophy. The novelty of our approach lies in developing (a) real world systems that themselves follow and observe-analyze-adapt paradigm at an operational level, and (b) real world applications that are able to capture application concepts and execution at multiple levels using events as a fundamental abstraction and event-based semantics as a unifying framework.