QCM Hardware Technology for Edge Computing

For applications requiring ultra-fast on-board monitoring of critical components, systems or infrastructures, Ontonix offers a Complexity And Health Monitoring System (CAHMS) – a device that can be mounted on cars, aircraft, submarines, helicopters or missiles or on any mission-critical equipment and deliver early warning signals of imminent systemic malfunctions. The CAHMS can be interfaced to a CAN bus or any source of streaming data. It can anticipate anomalies and problems, not just signal their presence.

The device is based on FPGA technology and has been developed by Ontonix and SAIC in the framework of a project funded by the US Department of Defense Executive Agent for Printed Circuit Board and Interconnect Technology and the NSWC (Naval Surface Warfare Center).

Each device is based on a Digilent ZedBoard, which includes a microprocessor that manages all data Input/Output as well the complexity chip itself. Dimensions: 16 x 16 cm. A smaller version of the board is also available: dimensions 16 x 8 cm.

Thanks to the CAHMS it possible to monitor complexity in real time and to identify, anticipate and diagnose problems in all those mission-critical systems which depend on increasingly intricate software environments and electronics:

automobiles
aircraft
trains
turbines, compressors
military equipment, mobile platforms (tanks, helicopters, submarines, aircraft carriers, missiles, armored vehicles, drones, etc.)
etc.

When integrated with a software/electronics system of an automobile or an aircraft, the CAHMS can detect systemic problems, i.e. those ‘nasty’ problems which occur in large and complex systems or networks and which conventional technology can acknowledge only after the fact. In addition, the system is able to deliver information that is key in the context of intelligent Preventive Maintenance.

The salient characteristic of the device is that it is able to monitor systems from a holistic, multi-discipline perspective, analyzing and combining information from heterogeneous data channels processing directly raw data. This is done without any training or Machine Learning. Yes, anomalies are detected without resorting to Machine Learning and indicating them the first time they occur.

The system is interfaced to a CAN bus.

The following real-time output is produced: