Host: Prof Takahide Mizuno
Kanako, Chris and I went to meet Prof Mizuno who is an associate professor at the Institute of Space & Astronautical Science. The Institute is located in a smart new building complex in Kanagawa Prefecture, not far from Tokyo, and has some impressive-looking moon buggies in the foyer.
According to the brochure Prof Mizuno gave us, as of January 2001 there were 43 full professors and 47 associate professors at the Institute, plus an additional 28 visiting professors, and 58 research associates. They are supported by 149 administrative and technical support staff, with an additional "several dozen non-regular administration staff and technicians". There are also 183 graduate students, and 11 "postdoctoral staff". The brochure does not mention any undergraduate students, but Prof Mizuno told us that there were some. He also mentioned that their postgraduate students have first degrees from various different universities.
We were joined by two post-doc research associates, two postgraduate students and one undergraduate student, with three more students joining in during the presentations.
About the Index project
Prof Mizuno told us about the Index (Innovative technology demonstration experiment) project, in which a team of five professors, eight research associates, eight students and three technicians have been designing a satellite. The project manager is Prof Saito. The flight model is to be built in the summer of 2002.
Prof Mizuno told us that the Index satellite is "very small" (50x50x50 cm3, weighing 50kg) and, at ¥0.4 billion (a little over £2 million) including launch costs, "cheap". By way of a contrast, he told us that the typical cost of satellites at the Institute is about ¥15 billion.
The Index satellite was due to be launched in June 2002, but Prof Mizuno said that it will have to be delayed because it is being launched in conjunction with a larger satellite that has had its launch date postponed for financial reasons. (They will not delay assembly though.) It was originally supposed to have been a two-year project; the fact that it has turned out to take much longer is unfortunate for those students who will have completed their period of study without having had the satisfaction of seeing the satellite launched.
Prof Mizuno told us that the satellite control system will run on an on-board computer that has three 32-bit RISC processors, and that they are using SOI (Silicon-On-Insulator) for the ROM on the on-board computer because of its inherent radiation hardness. Once in orbit, the main scientific mission of the satellite is intended to be observation of the aurora; its purpose is primarily demonstration of new technology and aurora observation, with teaching being a secondary aim. Its lifetime, after launching, is expected to be "more than three months".
Presentations from Research Associates/Students
Following Prof Mizuno’s overview of the Index project, we were treated to presentations from three of the research associates/students. The first presenter was Dr Fukuda, one of the research associates. He explained the integrated control system that is hosted on an onboard 32-bit high speed RISC processor – a commercial product manufactured by Hitachi (called SH3?). He told us that the system encompasses positioning and altitude control, command and data handling, tasks associated with the satellite’s scientific mission, and telemetry. The system is based on an 8Hz interruption scheme. He explained that normally there would be a separate processor dedicated to each of the major control tasks, but they have been combined into one within the Index satellite in order to keep the size and weight down. The processor works in conjunction with a radiation-hardened FPGA chip from Actel.
The second presenter – Dr Sakai, also a research associate – presented his thoughts about the potential usefulness of Handel-C for control engineering. He had clearly prepared this specially for our visit, although what he had to say wasn’t specific to Handel-C. He drew a schematic block diagram of a typical (generic) system with feedback, where sensors feed back data to a controller that in turn exercises control over the system via actuators. He pointed out that much of control theory involves matrix manipulation, and with that in mind he views an FPGA as a "flexible vector calculator" that could form the heart of a controller in a system such as the one he had drawn. In particular, the motion controller within the Index satellite uses matrix manipulation extensively.
Dr Sakai suggested that the additional computational power that an FPGA would introduce could be harnessed in one of three ways. Firstly, it could enable manipulation of larger matrices ... but nobody knows how to "tune" very large matrices, so this is of little benefit in practice. Secondly, it could enable faster matrix manipulation ... but there’s little point unless the speeds of the sensors and actuators can be increased to take advantage of this. Thirdly, he suggested a novel approach that he said he felt quite excited about involving reconfiguring the FPGA after launch. If my understanding was correct, his suggestion was that the controller process itself could be fine-tuned by a more powerful computer on the ground that would calculate and upload successively refined control algorithms to the FPGA. In other words, it would be an adaptive system involving a powerful computing platform on the ground inter-communicating with the satellite, and with periodic reconfiguration of the on-board FPGA. He added that this was "just a dream" at this point.
The third presenter was Mr Takahara - an undergraduate student. He gave a brief overview of how a more reliable system could be created by having three identical sub-systems and a voting mechanism. Each sub-system would contain both a CPU and an "SRAM type FPGA". This wouldn’t provide any protection whatsoever from design errors (because any design errors would be present in each of the sub-systems), but it would provide a degree of protection against corruption due to the presence of heavy ions in space. The idea is that if one of the FPGAs was found to have been corrupted it could simply be re-configured.
Miscellaneous
Prof Mizuno said that they are interested in evaluating the suitability of Handel-C/DK-1 for designing a new FPGA for the successor to Index. He added that "Index-2" will almost certainly be substantially different from the current Index satellite, with new sensors and actuators, warranting a new FPGA design. The ideas that Dr Sakai and Mr Takahara had described were also intended (potentially) for Index-2 rather than Index. (Since the aim is to provide students and junior researchers with hands-on experience of satellite design, presumably reuse of existing components isn’t as desirable as it would normally be.)
He also said, not surprisingly, that relatively few of the graduates from the Institute of Space and Astronautical Science subsequently find employment within the aerospace industry. Instead they disperse into various areas of Japanese society.
Acknowledgement
I am indebted to Prof Mizuno for his feedback on this report.
|
|