White Paper

Intel-Safety-Certifiable-Computing-Tomorrow-Avionics-Whitepaper

Issue link: https://read.uberflip.com/i/1382624

Contents of this Issue

Navigation

Page 1 of 6

WHITE PAPER Evaluating Safety-Certifiable Computing For Tomorrow 's Avionics mrcy.com 2 mrcy.com 2 Next-generation autonomous aircraft face many challenges. Urban Air Mobility (UAM) platforms are converging multiple workloads on centralized computing to save power, weight and cost. Ar tificial intelligence (AI) and machine learning (ML) applications enabling autonomous aircraft demand vastly increased processing per formance but must also be cer tified safe. The longtime common choice for safety-cer tifiable processing, PowerPC® architecture, has reached the end of its product life cycle and system architects need to find new long-term processing solutions. Facing these rapidly emerging challenges, Mercur y Systems initiated a study to examine safety-critical application needs and evaluate the suitability of ARM and Intel processors for next-generation platforms. AVIONIC TRENDS IMPACTING CERTIFIABLE PROCESSING Through its investigation, Mercur y identified five unique areas driving avionic computing requirements and evaluated current available safety- cer tifiable processors against these needs. 1. Digital Convergence – Aircraft have traditionally dedicated one subsystem for each application due to processing and footprint restraints. This closed architecture approach complicates integration, creates multiple vendor locks and makes upgrades difficult. Future avionic systems architectures will need to run applications on a single, powerful avionics server with software-defined functionality to keep pace with growing data loads and evolving applications. 2. Complex Application Capabilities – In the past, applications were constrained because of legacy processors that could not scale to meet growing performance requirements. Users are demanding more functionality and capabilities, driving the need for complex, processing-intensive applications 3. Autonomous Flight – The recently launched Urban Air Mobility market needs fully autonomous flight computers, while defense platforms are looking for semi-autonomous capabilities to offload pilot workloads. These evolving needs will likely require a combination of AI/ML techniques supported by heterogeneous processing (FPGA, CPUs, and GPUs). 4. Increased Demand for Certified Graphics – High-resolution displays, the demand for 10- bit sensors and fused graphics a need certified graphics processing units (GPUs), either as discrete components or integrated within the CPU. 5. Safe and Secure Processing in Avionics – While safety is important, there is a growing concern over cyberattacks and compromised supply chains in critical avionic subsystems because they could jeopardize product integrity and systems safety assurance. Europe has long required safety certification for most of its defense platforms and now many U.S.-based defense platforms are following suit. As the demand for safety increases globally, so does the complexity and performance requirements of the computing systems powering these critical applications. Converging Functions into Avionics Mission Common Servers. Outdated technology is risky. It can render progressively compute- heavy AI systems ineffective. Using a single, powerful mission computer for multiple functions supports rapid modernization of mission-critical systems and accelerates AI deployment through design flexibility and reduced complexity. Flight Control Converged Mission Computer Graphics Display Communications Navigation

Articles in this issue

Links on this page

view archives of White Paper - Intel-Safety-Certifiable-Computing-Tomorrow-Avionics-Whitepaper