FPGAs in Space Communication for Dummies

Michał Kuklewski, Thales Alenia Space

Due to their versatility and configurability, Field Programmable Gate Arrays (FPGAs) continue to be a very attractive choice in many places instead of purpose build ASICs. Space applications – in low or high orbit or even on other planets – are one of them. Depending on mission conditions or requirements, there is the possibility of using commercial off-the-shelf (COTS) devices in contrast to radiation-hardened FPGAs with built-in protection by technology. The use of any of these has associated implications and requires appropriate protection to assure reliability and achieve its function. 

As an example, the application is terminals for space communication – either between a ground station and a satellite or between satellites.  To achieve it, space agencies are following CCSDS standards, thus enabling the interconnection of different systems, while minimizing risks during missions, maximizing success rate and achieving interoperability. 

This presentation features a compilation of information from the ECSS standards and handbooks for protecting FPGAs against the effects of space radiation and an introductory description of CCSDS communication standards that allow communication with satellites. 

Mars – the most fascinating target for the space exploration
 – the technological challenge

Piotr Orleański DSc, CBK PAN Warszawa

The lecture will present the summary of the different space missions to Mars with the special attention on the technological challenges (environment conditions, available power, communication issues, components reliability). The more detailed information will be given on the three missions (Mars Express, ExoMars and InSight) where Polish engineers have been involved in design and manufacturing of the space instruments. 

Sending humans to Mars: a beginner’s guide

Jakub Hajkuś, To Jakiś Kosmos

So, you’re about to go to Mars? That’s great! Pack your toothbrush and lunch box. Your spacecraft isn’t ready? Oh, You don’t have one yet. Let’s try to do something about it. It’s not easy to get from Earth to Mars, but I’ll walk you through the process. Leaving our blue planet, cruising and landing on Mars require creativity. I will introduce you to a few options from which you can choose. Better clear your schedule as the journey usually takes several months. Warm clothes and sunscreen are a must. Being in space or on Mars can hardly be called a perfect holiday. I almost forgot! Consult with your doctor about whether you may be exposed to strong cosmic rays. Don’t worry, you are in good hands!

The Hitchhiker’s Guide to the FPGAs.
How to make FPGAs work in space

Adam Milik DSc, ALDEC

Before our electronic system leaves the ground, several problems need to be solved and tested.  Space radiation is one of the problems that are invisible at the ground level. The problem of single event upset will be addressed, its influence on FPGA architecture, modeling concepts, and methods of overcoming. The next section addresses the problem of power estimation in digital systems. The power consumption model construction will be shown and the implementation of the multiple clock domains system. The last section focuses on the ideas of algorithm mapping to hardware structures and efficient resource utilization. We will examine the resource sharing concept and its hardware effectiveness, building operation schedule, and its transformations for effective implementation. 

Using FPGAs for pre-silicon validation

Paweł Dowgiałło, INTEL

Every silicon development and fabrication is subject to the risk of major design defects resulting in a significant cost of fixing them. The risk gets higher with design complexity. High complexity increases time-to-market which, from an economical perspective, must be as small as possible to maximize commercial success. This trade-off is being addressed by companies in multiple ways, one type of recently most effective techniques is emulation and prototyping. In this area, FPGAs play a fundamental role thanks to their ability to implement and run silicon RTL with cycle-accuracy and within a relatively short time. The speech will cover the gains of applying FPGAs to address silicon validation risks, outline the process and highlight major difficulties and challenges. 

Hardware design performance insights at up & running FPGA

Andrzej Stasiak PhD, INTEL

There are many tools and methods to estimate and statically calculate the design performance before we move the RTL to FPGA. Once we implement the project and get an expected benchmark number via simulation/estimation, it seems the project is done. The problem may arise and become significant when our design due to internal / IO / other dynamics, does not meet set performance minimum threshold while runtime. The architecture/design itself then needs deep investigation while the runtime. This article discusses methods and techniques, including data collection and further offline analysis, to narrow down the design performance bottlenecks. 

Vendor Independent FPGA Implementation
for Safety-Critical Designs

Rakesh Jain, SIEMENS

Today’s designers have many choices – between ASIC/FPGA and multiple FPGA vendors, such as NanoXplore, Xilinx, Microchip and Intel. Siemens EDA has a complete FPGA vendor-independent flow from Catapult HLS, Precision FPGA Synthesis and FormalPro LEC/Questa verification. For safety-critical and high-reliability applications, Precision Hi-Rel offers industry’s most comprehensive SEE (Single Event Effects) mitigation strategies – TMR, safe FSM, ECC RAMs, etc., and, integration with provides assurance that synthesis-based mitigated design is functionally equivalent to the RTL, ensuring DO-254 certification. 


Integrated Workflows for Deploying
Deep Neural Networks on FPGAs

Dimitri Hamidi, MathWorks

Deep learning on FPGAs is playing an increasing role across a growing number of systems and applications due to FPGAs’ unique attributes of flexibility, high throughput, low latency, and per-watt performance. Nevertheless, the high development effort and the need for extensive hardware knowledge might become a prohibiting factor when targeting deep learning to FPGAs.  

This presentation illustrates an integrated workflow that simplifies the task of design exploration and prototyping of deep neural networks on FPGAs for control design, signal, and image processing applications.  

It connects algorithm, system, and hardware teams and enables early collaboration while accelerating the development process, both in the concept phase and in the implementation phase.  

The workflow is centered around a customizable deep learning processor with synthesizable RTL code generation support. Moreover, deployment for prototyping does not require specific hardware skills.  

Prior to deployment, the user can rapidly prototype the custom deep learning network and bitstream by visualizing intermediate layer activation results and verifying prediction accuracy without target hardware since the emulation is supported. Furthermore, performance and resource utilization can be pre-estimated and can be later confirmed with synthesis results and on-board profiling data after rapid deployment. The user can further fine-tune the performance to meet system constraints by redesigning the networks, fixed point quantization, and by trading off hardware resource usage with latency.  

The modular deep learning processor is portable, customizable and can be integrated into a reference design via AXI interfaces. Moreover, the integration process can be automated. 

We describe the architecture of this processor in detail and show how it can be customized and quickly deployed for deep learning inference on FPGA boards.   

The Space around us – You didn’t know that about smartphones

Tomasz Rożek PhD, Nauka To Lubię

Have you ever wondered what impact space technologies have on our everyday lives? When you answer the phone or fry eggs and bacon, you won’t even think that all this is possible thanks to discoveries made in super advanced laboratories. And at the beginning, these technologies were not invented to fry eggs … 

Many of the technologies that were created to explore the Cosmos are now used to improve life on the planet Earth. What would an average flat look like without them? A bit empty. Because we do not even realize how many of the advanced inventions that we use today to perform completely mundane activities. That’s why it’s so easy to amaze us with the information that baby diapers and a Gore-Tex jacket come from outer space. One of the most cosmic inventions are smartphones, without which most of us today cannot imagine life. If not for space missions, smartphones would probably not be there. The technologies used for their production come from space missions.