Real time SAR‐data processing: Parallelism and granularity illusions and possibilities

The problem of precision processing of SAR data in real time is here discussed, considering both a new performance evaluation model for structured algorithms and effective computational schemes of the SAR data compression process, in order to assess conditions of effective real time processing possibilities in this performance demanding application. Our analysis leads us to propose a fine grain architecture as a reasonable solution. Rigorous performance estimate is also provided for it. The single lowermost-level processing element "ad hoc" designed by us, has been yet implemented in VLSI technology and is under test; it is suitable to work in very high granularity systems of bit-sequential type (as "connectionist" architectures). Interconnection and data allocation policies, as well as possible process decomposition are also discussed.