Search

A proposal for a new space mission for a compact and efficient telecommunications satellite, compatible with a launch of the Vega space transport system, is illustrated. The idea of a Vega compatible telecommunications satellite mission falls within the lines of re-launch of satellite telecommunications programs by the Italian Space Agency with missions already under development and with a view to enhance and strengthen the domestic industry and the systemic capabilities already pride of the country. In addition, the synergy with the Vega launcher, and its potential developments, encourages developments of great interest for the role of Europe in the launch activities, extending its potentials. The proposed satellite has in fact a mass and a volume lower than those typical of telecommunication satellites, but with performance if not equal, similar to the latter. The mission seizes the opportunities offered by the Vega launch vector and the performance of electric propulsion, characterized by low thrusts and small masses of propellant.

Although new generations of small satellites have power imaging capabilities that make them comparable with large satellites; they suffer from a limited downlink transmission. The use of data compression on-board the spacecraft will allow the reduction of the size of images onboard and shorten the transmission time. The adopted approach to lossless image compression employs predictive neural networks, integer wavelet transforms and Peano-Hilbert scan. The benchmark results show that the compression ratios obtained with the neural network-based method combined with a Peano Hilbert scan are higher than those obtained with JPEG2000 and the Rice algorithm. Ways of speeding up the processing time of the proposed algorithm have been undertaken. It has been shown that the number of neural network sets and the size of image tiles affect the processing speed of the neural network predictor. A 3-set configuration represents the best tradeoff in terms of performance and complexity. A hardware implementation targeted at FPGA platform has been studied and simulated using a VHDL model. The concept of our proposed neural compressor is based on multiple neural processors performing the compression of a certain number of image tiles in parallel. A decision support scheme was introduced, which makes the compression adaptive to the image zero-order entropy and aims to minimize the on-board memory usage. The concept behind the adaptive scheme is to allow an individual NN processor to execute a 2-set prediction instead of a 3-set one provided the resulting loss in performance is tolerable. A solution to incorporating the novel compression technique in an on-board satellite image system is outlined. A new hardware system (SSONICS) is proposed realising lossless image compression in an independent decentralized way without using the resources of the on-board computer fully. The system can easily be extended to comprise additional image-processing techniques allowing decompression and recompression without changing the internal structure of the hardware completely. This novel system can facilitate a faster way of data transmission to ground terminals and therefore can offset the data link problem.