Germany | Bayern

Zurück zur Suche

Internationale Partnersuche

Forschung & Entwicklung Anfrage

Horizon 2020-MSCA-IF: Research opportunity in the field of high-performance data processing algorithms and artificial intelligence for embedded computing

Country of Origin: Switzerland
Reference Number: RDCH20200707001
Publication Date: 14 July 2020

Summary

Swiss company develops fast data processing with lower power consumption for operators in areas such as space, robotics and mobility. Computer science researcher sought (PhD/experienced) within an Individual Fellowship of a Marie-Skłodowska-Curie action for a research cooperation.

Description

There is a growing need to perform tasks 'on the edge' (at the point where the data is produced) as opposed to 'the cloud', this is especially important to sectors like space where processing capabilities and power are extremely limited, for example in the case of earth observation CubeSats. Applications to be performed on the edge can be divided in two groups: 

1. Data processing algorithms such as Kalman Filter, Fast Fourier Transform (FFT), etc
2. Deep Neural Networks (DNN) on the edge

These applications have a major drawback of needing significant on-board computer power, and therefore can compromise the overall power of the satellite.

Several solutions have been studied for edge data processing:

- The use of Field Programmable Gate Arrays (FPGA), i.e., programmable hardware, can reduce power consumption and increase data processing, however high complexity in programming is the main drawback of FPGA.

- Graphics Processing Units (GPU) are probably the fastest data processing processors today. However, large power consumption, thermal load and performance bottlenecks in data transfer to GPU memory reduce their appeal in space applications.

- Software solutions in the host computer are the most attractive solution due their programming simplicity and relatively good performance. However, power consumption is high and data processing performance is often not fast enough.

These solutions, without a high-performance data processing algorithm accompanying them, are unable to meet power budget and/or data processing requirements.

In order to perform Artificial Intelligence (AI) or other advanced algorithms on-board, there are standard techniques and frameworks, well established both in industry and academia. These types of techniques are oriented to optimise the use of computational resources for specific mathematical operations like matrix multiplications.

State-of-the-art research, however, is showing that there are new approaches to algorithmic implementations for specialised hardware like FPGA, GPU and Neural Processing Units (NPU) that can perform better than standard techniques while using less energy consumption.

The goal of this project is research deeper into these new techniques so that they can be applied to advanced edge applications as mentioned above. Achieving such a goal, will give our company a big competitive advantage for sectors like Space, Robotics and automotive.

The company is looking for a computer science researcher (PhD/experienced) within an Individual Fellowship of a Marie-Skłodowska-Curie action for a research cooperation.

The deadline for EOI's is 31 Jul 2020, the deadline of the call is 09 Sep 2020 and the project duration is 144 weeks.

Requested partner

The candidates should meet the following criteria:

• PhD. in computer science (specifically C/C++, embedded
systems, FPGA, GPU, NPU) or 4 years research experience
in the field
• Open minded, creative, and thorough researcher
• Follow scientific methodology and attention to detail
• Proficient in English.
• Having previous publications in the field is not mandatory,
but it is a plus

The fields of research that the company is interested in are:

• Multi-thread programming
• Experience with FPGA, GPU or NPUs
• Vectorization tools OpenMP, TBB.
• Signal processing algorithms (Fast Fourier Transform,
Kalman filter, etc.)
• C/C++, MATLAB, VHDL, CUDA languages

The tasks to be performed are:

Thorough research on performance of several embedded computing scenarios including multiple processors, operating systems and algorithmic applications.

Cooperation offer ist closed for requests