Chimera

Chimera schedules task processing and uses a plug-in approach to connect multiple systems. Its modular design enables faster development times and reduced cost. Agility and cost reduction is achieved through:  

• Rapid and scalable open architecture mission system design and modernization
• Significant reduction in recurring engineering tasks
• Increased processing efficiency through intuitive data flow and manipulation to support dynamic resource use and management
• Self-managed data queues and scalability at a thread, process, or processor level to maximize flexibility
• Growing library of Chimera modules available for integration. Coding modules:
  • Remove the need for developers to write and maintain complex code
  • Enable simplified units of code that concentrate on critical processing
  • Do not require Chimera source code changes
  • Support independent module verification and built-in type protection for an extra layer of integration safety and effective unit testing

Chimera is designed for use in real world mission environments and thrives in image processing applications that require high-demand, processing-intensive and resource limited missions. This approach fosters a rapid development environment and reduces engineering task load.

Latency considerations are important for any image processing architecture. Chimera plugins are written to use available technologies such as NVIDIA CUDA to reduce latency. Docker, with the NVIDIA Docker extension, allows Chimera to interface directly with the graphics processing unit (GPU). Chimera was designed to accommodate scenarios where a user may have an environment with GPU limitations. In such a situation, field programmable gate arrays (FPGAs) are made accessible to Chimera through Docker. A Chimera plugin places bytes into the FPGA memory space and another Chimera plugin retrieves the results.

Chimera’s image processor expands into a distributed computing environment. Modules are added to each Chimera instance to pass large imagery and associated metadata between instances. Each Chimera can be “right-sized” to perform whatever is needed on each node.

Chimera supports swift redeployment of mission processing capability through containerization via Docker. The use of containerization provides several benefits, including keeping applications separate from infrastructure.

Dependencies of the application are wrapped into the container and health checks provide automatic mitigation of container or hardware failures. Container-tagging capability permits previous software versions to be available in the event that a user needs to perform a software fall back. The use of containers provides an environment that is safe for development and does not affect the entire code base. This functionality allows for Chimera instances to be managed without user management.

The Chimera software is composed of Chimera, Lycia and their dependencies.

• Chimera modules are coded in C++ 17
• Chimera will run on any POSIX compliant modern operating system
• Chimera provides a framework for developing modules that permits Lycia the capability to load and execute dynamic libraries
• Kubernetes and Docker supported containers
• Lycia is offered as an associated executable and can be used to instantiate a Chimera environment, read configuration files and contain the Chimera runtime
• Forge is the Chimera development environment that can be tailored to specific customer needs
• XML and JSON libraries provide support for parsing configuration files
• Time provides formatting for timestamps