Global Ambient Intelligence Network (GAIN)
GAIN is designed to enable the rapid implementation of sophisticated, distributed systems for homeland security and many other applications. Wireless, fibered or wired GAIN nodes can be configured for sensing /actuation, embedded processing and storage and enable sophisticated user interfaces. GNAT (Global Networks Academic Test) is an FPGA-based concept demonstration. Working with industry partners such as the MathWorks and Intel, we are developing tools and an integrated design methodology to enable high level design, simulation of nodes and overall GAIN analogous to the design, simulation and verification of CMOS VLSI using CAD. Our work has support from the IEEE Computer Society Technical Committees on Design Automation (DATC), Semiconductor Research Corporation (SRC) and the Microelectronics Advanced Research Corporation (MARCO).
The challenge of insuring the dependability of our critical infrastructure requires understanding their range of natural behaviors and identifying anomalies. Amazonian flows of data course through our transportation, power, financial and telecommunications networks. Filtering, analyzing, correlating and interpreting these flows to extract an anomalous pattern of behavior that might indicate an incipient failure, or a developing attack, requires a new approach that blends what electronic computation and human intellect each do best. In effect, the network is "aware" of its own range of normal or near normal states and even the regular state transitions. When a significant deviation from what is expected occurs, the network can identify the nodes where the deviation has occurred and notify the human network management.
The late Albert Szent-Gyorgyi, 1937 Nobel Laureate in Medicine, defined discovery as "...an accident meeting a prepared mind [and]... in seeing what everyone else has seen and thinking what no one else has thought." The Advisory Group to the European Community's Information Society Technology Program (ISTAG) defines Ambient Intelligence as "the convergence of ubiquitous computing, ubiquitous communication, and interfaces adapting to the user." We propose the establishment of the Global Ambient Intelligence Network (GAIN) to: disseminate, promote and foster the development of ambient intelligence technologies that are key to discovering anomalous behavior in our critical infrastructures.
The GAIN will help enhance the dependability of our vital systems and networks by linking sensor, reconfigurable system-on-chip processor, ad hoc wireless network communications and interactive 3D data visualization core technology development, formal and informal education, development and implementation of standards for interfaces and data structures with specific applications.
GAIN will bind research and development, educational organizations and industrial suppliers with private and governmental sector operators of our critical infrastructure. GAIN arrays with distributed processing and storage can recognize and report anomalies that are systematic deviations from the complex natural variability of the environment (a form of "self awareness").
Our research work involves developing an integrated design methodology for Programmable-System-On-a-Chip (PSoC) based ad hoc sensor arrays. At UNH, we have taught and employed the traditional Microelectronic Design Methodology for many years. This high-level chip design methodology (e.g., Gajski Diagram) is being extended to PSoCs by using libraries of simple well-characterized standard elements (IP). Our team has implemented a small-scale GAIN network, that we call the Global Network Academic Test (GNAT). GNAT's board-level nodes, based on off-the-shelf Field Programmable Gate Arrays (FPGA, e.g., Xilinx Virtex series) can incorporate multiple sensors, signal processing, local storage and node to node communications.
We propose to extend GNAT in two directions: down inside the chip, to enable nodes integrated at the chip-level; and upward to the Global-scale, with pre-configured and ad-hoc distributed constellations of networks of reconfigurable nodes. Eventually, this process can be extended to constellations of arrays of sophisticated chip-level versions of GNAT nodes. Nodes can include: multiple general and specialized processor cores (similar to the Intel Terascale concept chip with 80 cores); integrated MEMS and nano-based sensors and actuators, on board mesh-type communications networks; local storage; external high performance sensor and communications ports.
These constellations of networks can include fixed pre-configured and pre-deployed arrays and ad-hoc arrays deployed as required, by people, animals (e.g . dolphins), and manned or unmanned, remotely operated or autonomous vehicles. The minimally or fully-capable nodes can be deployed by mobile robotic vehicles (space, air, ground, water, (surface and submerged)) to enable the rapid implementation of sophisticated distributed sensor arrays.
We have nicknamed this process "Johnny Sensorseed" after John Chapman who planted apple seeds, the eponymous Johnny Appleseed. Our robot functions as Johnny in distributing minimally capable sensor nodes (the Sensorseeds). However, just as Chapman also planted seedlings and saplings, we can also incorporate nodes with more processing and storage (Sensorseedlings) and /or fully capable nodes with long-range wireless connectivity (Sensorsaplings).
A prime application of the Johnny Sensorseed concept is associated with Project PLUTO (Platform Laboratory for Underground and Tunnel Observations) that we are pursuing with the Central Mining Institute of Katowice Poland. PLUTO aims to develop GAIN sensorseed and mobile robotic technology specifically for use in monitoring mines and tunnels and assisting rescue teams at the site of an incident.
We are pursuing a phased approach, focusing first on an integrated design implementation methodology with appropriate metrics and assessment procedures. We have industry partners including MathWorks and Intel at the core technology level; Microstrain and iRobot at the node and platform level and SpaceFlight Systems and Profit Tools at the applications level.