FPGAs in Space; Low-Power Signal Processing Capacity for Nano-Satellites

By many, nano-satellites are seen as a promising divergence from the conservative design philosophies that makes current aerospace solutions excessively expensive. As a result of the price-barrier, very few have the economic muscle or willingness needed to realize and experiment with space-based ideas that potentially could benefit us all.

Nano-satellites changes this by adhering to a strict KISS design philosophy, using COTS components and most importantly by constraining the weight to 1-10 kg. As a result, the price-barrier is lowered, thus increasing accessibility to aerospace technology for entrepreneurs.

Despite the obvious benefits, one of the major challenges with nano-satellites is that the efficiency of solar-cell technology have not scaled at the same rate as the ratio between power consumption and surface area of the satellites; in fact solar-cell efficiency have stayed almost constant. Hence, power is a serious constraint in modern nano-satellite systems, and limits the capabilities of the technology.

One of the big consumers of energy on satellites is the digital signal processing associated with mission payloads; common examples being image or radio signal processing. Hence, the focus of this project is to research possible avenues that enables flexible and low-power signal processing while guaranteeing real-time constraints. The insight followed in the research is that signal processing algorithms often have an inherent tolerance towards noise and computational errors, which can be used as leverage to reduce power. More specifically, we intent to investigate voltage over-scaling of field programmable gate arrays (FPGAs), which is a reconfigurable signal processing architecture. Voltage scaling reduces power, but also slows the circuits which results in timing errors. Hence, a major part of the research involves quantifying and modelling the effects of voltage scaling in terms of timing errors. Finally, we aim at proposing methods that can tolerate such errors while producing acceptable signal processing outputs and investigate the signal quality and power trade-off hereof.

The project is executed in cooperation with and under the supervision of GomSpace who will also provide testing access to their experimental satellite platform GOMX, due to launch in Q1 2013.

Funding is provided by Vækstforum through Center for TeleInfrastruktur (CTIF) at Aalborg University and GomSpace APS.