Conquering Big Data means: get the code to the data. The Big Data Analytics engine of PublicaMundi, rasdaman, now has been selected to drive this to the extreme. It will be installed on board a satellite so that geo analytics can be performed right when the images are acquired.
On the 16th of January the European Space Agency (ESA) has confirmed selection of a Big Data engine, developed at Jacobs University Bremen in collaboration with a small hi-tech company, to go on board of a satellite. Goal is to demonstrate how satellites can process data on board and participate in planet-wide analytics networks.
Satellite imagery is indispensable for rapid response to disasters such as floods and wildfires, but also for everyday’s governmental tasks, like land use monitoring: who is farming what and how much? However, data from a satellite take a long journey: when the satellite is flying over a ground station, it attempts to download as much of the imagery taken as possible. That is passed on to some data center which undertakes a lot of processing, puts the resulting products into its archive, and eventually publishes them. Not always is it easy for users to find the needle, that is: those data they are interested in, from the resulting haystack of data.
Researchers at Jacobs University want to change this now, allowing regular users to get data directly from the source. To this end, this image source, the satellite, will be instrumented to allow flexible processing and filtering of data on board. This is accomplished by the rasdaman system (“raster data manager”), an innovative analytics engine specialized on science and engineering data which is already in operation internationally, among others at NASA Ames and ESA. This effort is not just a one-time endeavour: insights gained can help to save substantial amounts of electrical energy in computers on the planet in future.
In ESA’s OPS-SAT project, a so-called CubeSat will be launched into an earth orbit. A CubeSat is a small, low-cost satellite which can be quickly and flexibly assembled for different configurations and tasks. CubeSats have been invented by US universities to overcome the high costs and time it takes to launch satellites, effectively hindering them from doing space research. While traditional remote sensing satellites cost hundreds of millions of Euros and have development times of a decade and more, CubeSats can be flexibly adapted to new experiments and launched with moderate costs. Today, the final specification of the satellite has been released for review by the Jacobs software engineers. Launch of OPS-SAT is foreseen for early 2016.
The OPS-SAT model hosts a Linux PC, camera, and GPS so that images and their position on earth can be recorded, processed, and transmitted to earth. Accessing the satellite will be done through open standards well known by geo service providers on Earth, so that no special programs are needed; any regular Web browser is sufficient. Therefore, the researchers see this as a contribution towards democratizing data access: on principle will allow anybody to access such a service via Internet – neither expensive receiving stations, nor huge data centers will be needed.