SERVICES & SYSTEMS INTEGRATION

EO Airborne Processing Modernization

S3

Summary

Challenge: 

  • More capability in the same size platform
  • QRC
  • Cost constrained
  • Legacy architecture
  • Interface with existing platform

Result: 

  • System built and deployed on time and within budget
  • High availability
  • Modern architecture

What we did: 

  • Mapped application
  • Developed a chassis, bp, aim
  • Rapid development/Delivery

Full Story

The customer had a challenge; their deployed, airborne, hardware-based solution had become obsolete. They needed to modernize to a new architecture to increase functionality, but didn’t have enough processing headroom in the existing system to add new capabilities. And, they wanted to do this with the existing platform and interfaces, limited funding, and a very aggressive schedule.

To meet the time and budget requirements, the Mercury team had to go from conceptual drawings to a complete, deployable solution in only six months. To do this required the team to leverage off-the-shelf elements wherever possible.  However, the specialized platform/environment interfaces did require substantial modifications to the chassis yielding a semicustom chassis that could pass qualification testing on a single pass.  This was done by using design techniques and experience from designing and building dozens of systems a year.

Although the customer had a deep understanding of legacy parallel systems; they had limited experience with modern compute architectures. To achieve their goal of adding new capabilities, the Mercury team worked with the customer to outline a new modern architecture for them, and how it would allow for more application headroom. The Mercury team mapped the customer’s application and overall dataflow to a completely different architecture -- systematically partitioning the problem onto the specified hardware to meet the requirements.

As mapping an application is a multifaceted and iterative problem; the Mercury team worked very closely with the customer to understand the problem constraints and interdependencies. We needed to determine how to move data into the system, execute algorithms and controls, and ultimately move information and data products out of the system while meeting hard real time requirements. In order to optimally map resources (i.e. CPUs, GPUs FPGAs and/or DSPs) to the algorithms and controls had to be understood from a compute, bandwidth and timing standpoint. This mapping and sizing process guaranteed that there would the system would not only meet its hard real time requirements but also have enough remaining headroom to add new capabilities.

Ultimately, the Mercury team delivered a working, deployable solution for the customer that exceeded their expectations. Mercury remapped the customer’s application onto a completely new hardware architecture, leave abundant room to add new capabilities, while remaining in the same platform with the same interfaces -- all within budget, and on time.

Get Information