7 SETI Program at the Medicina INAF Radioastronomy …
71
7.5 Conclusions
The tests described above showed very promising results that led us to continue the
development of the spectrometer line.
Both pilot projects, that have proven to be very useful in the design definition
phase, and the initial software development, will continue to be used for the same
purposes but also in applications other than SETI. The first final project will be optimized for narrowband scientific research (NEO, space debris, and so on). Its development seems to be very interesting for amateur and educational applications Finally,
the second final project will be optimized strictly for broadband (radio telescope IF)
SETI researches in piggyback operations.
Like in past experiences, SETI technical researches and related instruments proved
to be very useful also in other research contexts.
7.6 Possible Future Developments
Many future developments of the new spectrometer line both in the hardware
processing capabilities and in the application software algorithms and functions are
possible.
Regarding the first pilot project, we will carry out comparative evaluations on
different algorithms like KLT, HHT, Cognitive Radio, Agnostic Entropy, etc. under
the same input signal, both artificially generated and detected. Maybe we shall try
KLT optimizations on other computation accelerators.
For what concerns the second pilot project, the priority is raising its computing
performances. To get this result, we will follow two approaches. In the first approach,
we will try to use AMD GPUs with OpenCL to acquire experiences on an SDK that
is useful in the future if the prices of FPGA accelerators will drop. In the second
approach, we will set up a low-cost computer cluster (see Fig. 7.16) composed
of ARM-based boards (Raspberry PI, UDOO QUAD, Pine64, Nvidia Jetson, etc.).
ARM architectures, in the last years, showed promising results in HPC applications
[41] also because they are implemented with a better lithographic process than Intel
CPUs.
The first final project will probably be used in more specialized NEO observations,
optimizing their functionality for this purpose. Other exciting applications of this
spectrometer will be dedicated software versions for amateur and educational fields.
The second final project will be upgraded with newer GPUs (Nvidia Volta, Turing,
Volta, or Ampere family) to let users to use the entire DAQ available bandwidth and
to allow more sophisticated computations.
71
7.5 Conclusions
The tests described above showed very promising results that led us to continue the
development of the spectrometer line.
Both pilot projects, that have proven to be very useful in the design definition
phase, and the initial software development, will continue to be used for the same
purposes but also in applications other than SETI. The first final project will be optimized for narrowband scientific research (NEO, space debris, and so on). Its development seems to be very interesting for amateur and educational applications Finally,
the second final project will be optimized strictly for broadband (radio telescope IF)
SETI researches in piggyback operations.
Like in past experiences, SETI technical researches and related instruments proved
to be very useful also in other research contexts.
7.6 Possible Future Developments
Many future developments of the new spectrometer line both in the hardware
processing capabilities and in the application software algorithms and functions are
possible.
Regarding the first pilot project, we will carry out comparative evaluations on
different algorithms like KLT, HHT, Cognitive Radio, Agnostic Entropy, etc. under
the same input signal, both artificially generated and detected. Maybe we shall try
KLT optimizations on other computation accelerators.
For what concerns the second pilot project, the priority is raising its computing
performances. To get this result, we will follow two approaches. In the first approach,
we will try to use AMD GPUs with OpenCL to acquire experiences on an SDK that
is useful in the future if the prices of FPGA accelerators will drop. In the second
approach, we will set up a low-cost computer cluster (see Fig. 7.16) composed
of ARM-based boards (Raspberry PI, UDOO QUAD, Pine64, Nvidia Jetson, etc.).
ARM architectures, in the last years, showed promising results in HPC applications
[41] also because they are implemented with a better lithographic process than Intel
CPUs.
The first final project will probably be used in more specialized NEO observations,
optimizing their functionality for this purpose. Other exciting applications of this
spectrometer will be dedicated software versions for amateur and educational fields.
The second final project will be upgraded with newer GPUs (Nvidia Volta, Turing,
Volta, or Ampere family) to let users to use the entire DAQ available bandwidth and
to allow more sophisticated computations.