Instrumentation systems are mainly used to monitor and control processes, which may be either very simple in nature (e.g., controlling the temperature of a room) or very complex (e.g., preparing, fueling and launching rockets in space). In general, complex processes can be broken into a series of interrelated smaller processes. But depending on the complexity and the number of processes monitored, traditional centralized decisionmaking entities (e.g., control room computers) are heavily loaded with data. Most of the time, these processes are stable and “in control,” meaning that, although large amounts of data are being analyzed, usually it is not necessary for the centralized entity to make a decision. This equates to a large tangible cost (in manpower resources and/or computational time) to analyze the “nominal data.”

Creating decentralized systems, capable of monitoring and controlling smaller, simpler processes and sharing information among them, is highly desirable. These systems can utilize information to make determinations regarding the overall process stability. By decentralizing the decisionmaking process, the system becomes less complex and more cost-effective and aids in eliminating single points of failure in the process.

The following steps minimize the disadvantages of centralized decisionmaking:

Overall Process Importance Assessment
(Traditional Versus Proposed)

• Break down complex processes into simpler, smaller processes and their relationship rules with respect to the overall process (calculate process-importance weights for each simpler one).
• Define the basic knowledge rules to govern these simpler processes.
• Incorporate these basic knowledge rules at the sensor level (embed knowledge rules in sensors).
• Utilize the properties of wireless communication to share process knowledge and process information among the sensors, transducers, and controlling equipment (Wireless Sensor Network).
• Decentralize the process decisionmaking capabilities and locate these capabilities remotely at the site of the physical process.
• Using the relationship rules defined above, verify control of overall process by verifying control of simpler processes and by sharing process information among simpler processes.
• Monitor the health of associated sensors involved in the process. Validate sensors measurement outputs by applying process knowledge rules embedded in the sensors.
• Minimize raw data transfer by emphasizing transfer of process information (processed data at the sensor level) versus processing raw data, optimizing the communication bandwidth.
• Ultimately, decentralize overall process control by creating a distributed Smart Wireless Sensor Network. The control of the overall process is shared among the different process field entities and the centralized entity. Control of lower-level processes is performed at the process-specific location. Information is shared among entities. The centralized entity performs an overall supervisory function by receiving, processing, and coordinating process information, not raw data.

The proposed architecture will develop a Smart Wireless Sensor Network that utilizes the unique characteristics of wireless communication to monitor and control a specific process. Each sensor (from now on also referred as Remote Station [RS]) will have the following architecture:

• A wireless communication section composed of a low-power radio frequency (RF) transceiver system and a microcontroller.
• An analog section composed of sensor excitation and signal conditioning circuitry.
• A power supply section composed of battery, associated power management, and protection circuitry.
• A digital/control section composed of a microprocessor (microcomputer or similar), storage memory, and analog-to-digital conversion circuitry.
• Adaptive software algorithms to provide hardware configuration and control capability, as well as process-specific knowledge rules and smart monitoring capabilities.

A typical RS will contain not only information to monitor and control its individual functions but also knowledge related to other RS’s associated with the process being assigned. Process information will be downloaded into the process control microcontroller through the RF link.

On a typical operation scenario of the Smart Wireless Network System, RS x will be polled by the Base Station (BS) for information. RS x will acquire data from the sensor, verify the acquired data is within expected range or exceeds defined limits, and broadcast results back to the base stations. RS’s associated with RS x by their embedded knowledge information will decode the information sent by RS x and validate the received information by applying the process-specific rules. As soon as the BS completes a monitoring cycle, each RS has acquired information not only about its own sensor but also from the associated RS’s for that monitoring cycle. At that time, they have validated not only their own data by applying knowledge rules contained in memory but the information of other associated RS’s independently. By doing this, the proposed architecture can detect failures (like instrumentation failures) and flag them to the base station. The architecture can also independently assess the state of the assigned process (process in control or out of control) and validate it with the base station. Eventually, when enough confidence is built in the RS’s, process control could be delegated to the RS’s, relieving the base station of this responsibility.

Key accomplishments:

• 2002: Established system requirements, preliminary conceptual design, and software architecture. Implemented laboratory demonstration prototype.

Key milestones:

• 2003: Implement field prototype.

Contacts: J.M. Perotti (Jose.Perotti-1@ksc.nasa.gov), YA-D5-E, (321) 867-6746; and A.R. Lucena, YA-D5-E,(321) 867-6743
Participating Organization: Dynacs Inc. (Dr. P.J. Medelius and Dr. C.T. Mata)