Fehlertolerante Instrumentenrechner für kompakte Kameras auf Raumsonden (English shop)

Preview

Table of Contents, Datei (50 KB)
Extract, Datei (130 KB)

The aim in developing a computer for scientific instruments on space probes is to increase the scientific benefit of the instrument through adapted control or data pre-processing. For computers of camera instruments this means processing more images per unit of time and making them available for transmission to Earth. This throughput is essentially determined by the compression routines, which require a great deal of computing time owing to the high image quality demanded.

The transition from a traditional radiation-hard to a radiation-tolerant design opens up new avenues in the construction of a small instrument computer. Taking the high-quality wavelet-based JPEG2000 compression as an example of the data processing functions of a camera computer, it is shown how, under the boundary conditions of a space mission, the yield, the resource requirements and the fault tolerance (tolerance towards radiation-induced errors) of a DPU for a compact camera instrument can be optimised. In order to put the values of the radiation-tolerant approach into perspective, comparative implementations based on radiation-hard components are carried out. In the transition to a radiation-tolerant design for building a small instrument computer, the availability of radiation-tolerant FPGAs with large resources (system gates) is of great importance. This makes it possible to integrate within a single component a standard CPU for universal tasks (control) together with dedicated logic for a specific task (compression). Computationally intensive software routines can be replaced by hardware engines that can be executed quickly and, above all, in an energy-efficient manner, as is expedient for compression, for example.

To increase performance for compact cameras, a scalable IDA JPEG2000 Core was therefore designed which permits flexible partitioning into hardware and software. The scalable IDA JPEG2000 Core was integrated, among other things, together with a CPU in an FPGA within the radiation-tolerant instrument computer. For comparison purposes, an implementation of the IDA JPEG2000 Core within a radiation-hard DPU was carried out in parallel. Through the energy-optimised, hardware-supported compression, the throughput can be increased by up to a factor of 30 while maintaining high-quality compression results. For typical image generation rates of one image of size 1k x 1k x 14 bit every 1–2 s, a high scientific benefit of the instrument is achieved by means of online compression. Such an FPGA-based solution offers sufficient resources for mission-specific extensions. While the reliabilities achieved with the radiation-hard and the radiation-tolerant DPU are equivalent, the radiation-induced rate of reversible errors is considerably higher with the radiation-tolerant DPU. For typical environmental conditions this yields a calculated unavailability of 1 min per year. Since this moderate radiation sensitivity can be accepted in most missions, the radiation-tolerant solution (FPGA and further radiation-tolerant assemblies) achieves a compact instrument computer with high performance, energy-optimised operation and great flexibility.