Comments on "Draft Proposal for VLBI Standard Interface Specification, February 3, 1999"

by
Noriyuki Kawaguchi
National Astronomioal Observatory

(May 14, 1999)
(revised on June 2, 1999)

General Comments

A VLBI standard interface has long been expected by many VLBI astronomers who hope to use radio telescopes in foreign observatories being equipped with different types of recording system. It is very welcome for me to have the standard interface like as a proposal given in a document titled "Draft Proposal for VLBI Standard Interface Specification, February 3, 1999". In this short note I would like to make some technical comments on the draft. Also I would like to point out a fact that the standard interface does not give a perfect solution but leaves media incompatibility behind. We still need to use a "translator" or a "hybrid correlator". We should aware the importance of the media translation now partly performed in Mitaka VSOP correlator, National Astronomical Observatory, Japan. Easy translation should be taken into consideration on the standard recorder interface.

Scopes of targets

In considering the standard interface, following goals shall be achieved;
(1) Easy translation of a tape media The DOB output shall be connected to the DIB in the tape translation, not only to a correlator. The output connector of the DOB shall be mated to the input connector of the DIB.
(2) Tandem operation of two or more DTSs Two or more DTSs will be involved not only in correlation but also in data acquisition. Wide band data recording at a rate beyond 1 Gbps would be made on two or more recorders. Synchronized operation of the DTS to others is a matter of concern as much in the parallel recording as in correlation processing.
(3) Real time correlation Real time correlation would be a future important function of a reliable long-baseline interferometer, in which the data on a tape is always checked for the auto-correlation spectrum, cross correlated for monitoring the phase stability while the recording is carrying on. Single operation of DIB at an observatory, (and of DOB at a correlator) will be less important in future, though it would keep economical benefit on the recording (or playing back) cost. My preference is not on the separation of DIB and DOB boxes.
(4) Sub network correlation Future operation of a VLBI network connecting a number of telescopes over the world will be made in sub networking. In the correlation, dynamical resetting and offsetting of ROT might be important.
(5) Unified software support A complicated software control might cause difficulty in the operation of different types of DTS. The DTS shall be controlled under the FS9 Weld system following the VEX schedule file.

Basic DTS Configuration

Functionally or Physically the DTS is divided into two boxes as indicated in Figure 1. Usually the DTS of a "Data Recorder" has both functions of DIB and DOB, but by some economical reason it might be a good idea to separate the DTS to the DIB and the DOB for the use in VLBl data transport system.

The DIB and DOB is just the same ones as described in the draft proposal VSI specification except for adding timing signals I/O on both boxes. Even in case that the DTS is divided into the DIB and DOB, the both input and output functions of the timing signals is useful for various applications to be noted later.


Figure 1. Basic Configuration of the DTS (Data Transmission System)

Parallel recording of a high rate data

A high rate data transmission beyond 1-Gbps from an observatory site to a correlation center is a strong demand of a VLBI astronomer in their high sensitive observations. Parallel recording is one possible way to allow for such high rate data transmission. Figure 2 shows a connection of two or more DIBs. Timing signals from a master clock at an observatory is supplied to all DIBs with no time offset and multi-stream data from a data acquisition system are connected to each DIB in parallel. The same timing signals are probably used in a data acquisition terminal.

In this application synchronization of two or more DIBs is performed with a similar way commonly done in correlation processing.


Figure 2. Parallel recording of a high bit-rate data on two or more DIBs.

Synchronized reproducing in a correlation center

At a time of correlation processings, the data at a specific time on tapes is reproduced synchronously from two or more DOBs. Timing signals are successively to each DOBs. The signals are also transferred to a correlator for the use in synchronizing correlation parameters and the data reproduced.


Figure 3. Synchronized data reproducing at a correlation site.

Media Conversion


Figure 4 Media Conversion

Data Transmission

Bit stream rates on a fixed number of logical wires should not exceed 64 Mbits/sec. Reliable cable connection limits a physical number of pins within 50, and a physical number of data connections within 32. For simplicity in the connection, it is better to limit a number of connectors within 2. The aggregate bit rate thus shall be less than 2048 Mbits/sec. The data transmission in much higher bit rates with metal wires is NOT realistic. Optical transmission should be considered. The 2048-Mbits/sec transmission will be technically feasible in the next 10 years with a single pair of the DIB and DOB. By considering above limitations, the data transmission listed in the Table 1 is reasonable on the current technical conditions.

Table 1
Aggregated bit rate Number of Connectors Clock Rate Logical Number of Wires
64 Mbps 1 4 MHz 16
128 Mbps 1 8 MHz 16
256 Mbps 1 16 MHz 16
512 Mbps 1 32 MHz 16
1024 Mbps 2 32 MHz 32
2048 Mbps 2 64 MHz 32


June 3, 1999