Services

How SolidKosmos enables the use of COTS components in critical applications

At SolidKosmos, we specialise in enabling the reliable use of Commercial Off-The-Shelf (COTS) components in safety- and mission-critical embedded systems. Our products and engineering services help customers integrate, validate, and qualify COTS-based solutions through a combination of emulation-based fault injection, physical testing, and robust system-level design.

 

We support our clients across industries—space, automotive, defence, industrial, HPC, and medical—with:

Radiation and fault testing at international facilities (ions, protons, neutrons)

Development of custom Fault Detection, Isolation, and Recovery (FDIR) strategies

Design and execution of fault injection campaigns for functional verification

Dependability and safety analysis (FMECA, FTA, MTTF, availability)

Certification support for compliance with standards such as ECSS, ISO 26262, DO-254, IEC 61508

Enabling COTS components for new space applications

New space missions demand low-cost, high-performance solutions with fast development cycles. SolidKosmos enables the adoption of COTS components (e.g. FPGAs, MCUs, SoCs) in space missions by:

  • Conducting single event effect (SEE) characterisation and vulnerability assessments
  • Applying radiation-hardened-by-design (RHBD) techniques to commercial FPGAs
  • Providing real-time scrubbing, lockstep recovery, and health monitoring IPs
  • Supporting qualification processes and documentation for COTS component reuse in high-radiation environments

Dependable AI acceleration

AI accelerators are increasingly deployed in critical environments—from autonomous vehicles and satellites to edge devices in medical and industrial systems. SolidKosmos delivers fault-tolerant AI/ML acceleration through:

  • Custom FPGA-based AI designs using Xilinx FINN, Vitis AI, and other popular frameworks
  • Support for quantised and pruned neural networks
  • Integration of fault-tolerant logic into AI data paths (RHBD for AI)
  • Validation of AI models under fault injection and radiation conditions
  • Full-stack support from neural network to deployment platform (model → RTL → HW/SW integration)

Our solutions ensure that AI systems deployed in critical environments remain predictable and robust.