TBC-001
|
5.1.1.2.2
|
Data routing:
Since routing is done per frequency channel, the routing information
will contain the following for each frequency channel and phase bin
(TBC-001):
Channel ID & Phase bin ID
SDP destination node IP address
SDP destination node port
|
|
Q3 2017
|
TBC-002
|
5.1.1.4
|
Data types and constants (Visibilities):
Table 3: SKA1 SPEAD Items
Schedule Block ID: Schedule block ID used to uniquely associate
correlator data with a particular observation. This ID will be provided
to the CSP by ™. Size is SPEAD HeapAddressWidth (48bit) TBC-002
(SDP_REQ_INT-48)
|
|
Q3 2017
|
TBC-006
|
5.4.1.2.1
|
Data organisation (SPEAD Heaps and Streams:
Each heap will contain the following set of transient buffer data: all
voltage data samples from one receptor for a duration of 1s to 10s
TBC-006
|
|
Q3 2017
|
TBC-008
TBC-009
|
5.4.1.4
|
Data types and constants (Transient Buffer):
Table 18: SKA1 SPEAD Items
Buffer_data:
Type: ‘i’,n_adc_bits TBC-008
Shape: N_samples TBC-009
|
|
|
TBC-011
|
5.2.1.2.3
5.3.1.2.3
|
SDP will be the (FTP) server TBC-011 (at least one per receiving
node).
|
For SDP to be the FTP server poses a potential risk for SDP as it may
impact other real-time processing/services on the receiving node.
Requires analysis and/or prototyping to determine the best FTP tool to
use for the PSS and PST interfaces.
|
Q3 2017
|
TBC-012 |
5.2.1.2.1 |
If possible OCLDs will be divided into smaller
packets/files (on a time or candidate basis) and transferred to SDP as
soon as they are ready (TBC-012) … |
Comment from Robert
Lyon: “We do generally agree with this. However we note this definition
can be problematic. Whilst it makes sense to split the OLCD in this way,
it implies that CSP is able to send the data in this manner. For single
pulse search (SPOCLDs) this is likely the case. For pulsar search, it is
less clear. Whilst CSP may be able to send individual optimised
candidate cubes as they emerge from the signal processing pipeline, they
would then have to be cached by SDP to reform the OCLD. This is because
it is advantageous to process candidates collected during the same
observing session and beam together. It also has implications for the
data transfer protocol used. Is FTP the best protocol for such
transfers? Perhaps adding some detail and caveats to this statement will
help. i.e. it should be possible to send the OCLD or SPOCLD as either a
complete entity (file or object), or to send pieces of the OCLD/SPOCLD
(sub-components) as smaller packets/files as required via the data
transfer protocol used.” |
Q3 2017 |
TBC-013 |
5.3.1.6.3. |
Number of samples per file for dynamic spectrum
mode |
The number of samples per file is based on 10s of data at the
expected sample rate. This needs to be confirmed with PST. |
Q3
2017 |
TBC-014 |
6.4 |
Verification process is TBC |
Interface verification
including ITF activities to be discussed during EngCon17. |
Q3
2017 |