Versal ACAPs for space: silicon photonics and chiplets

An increasing number of satellite applications require intelligent in-orbit processing to extract value-added insights rather than clog precious RF downlinks with bandwidths of data for post-processing on the ground: some require autonomous, real-time decision making, e.g. space-debris retrieval or in-orbit servicing spacecraft outside of their ground-station coverage would not be able to receive late commands to initiate collision-avoidance manoeuvres, or space-domain awareness using multiple sensors followed by object detection and classification may require an immediate friend or foe decision. High-definition SAR imagery is increasingly generating huge amounts of Earth-observation data and in-orbit AI inference and neural networks allows for feature identification, scene segmentation and characterisation.

Space-grade FPGAs, ACAPs, MCUs with vector-processing engines and rad-tolerant AI accelerators optimised for linear algebra & neural networks, each offer certain advantages for intelligent on-board processing. Some applications require small, low-power, Edge-based solutions while others need the performance of a 130 W ACAP.

This presentation compares the design-in of FPGAs with AI accelerators, MCUs with vector-processing engines and Versal ACAPs, as each presents unique challenges for power consumption, d.c. plane drop and PDN analyses using Hyperlynx, optimisation of IO using IOPT, PCB stack design, component placement and track routing within Xpedition.