UNIQUE WORD DETECTION FOR LRPT
For 3 weeks or so Meteor transmitted its LRPT stream at 80kBits/sec using the unique word scheme as described
in the Metop LRPT mission specific interface document. This scheme inserts an extra 8 bit word every 72 bits
as the last step just before modulation. The byte is always 27HEX ( 0010 0111 ) and serves as an aid to allow
ground reception equipment to re-establish synchronisation with the original data stream. There are 3 immediate
problems after demodulation, QPSK mapping, synchronising the convolutional de-interleaver and symbol phasing
the viterbi decoder.
The phase-ambiguity introduced by the coherent demodulation process means that the data on the I and Q outputs
can appear either correctly, both inverted or I and Q swapped with one or the other inverted. In software, it is a
simple matter to search for the occurance of the of the 27h pattern in all the possible translated states. When a
unique word is found the phasing is noted and the translation is applied to the remaining 72 bits.
This code segment here shows my approach to the problem.
As for syncing the de-interleaver and viterbi decoder, the Metop spec states that the de-interleaver 36-way switch
starts in its initial position immediately after the unique word. Simply reset the de-interleaver at the start of each
recovered 72 bit block of data. The Metop spec also states that the first bit after the unique word will always be the
G1 symbol out of the G1/G2 symbol pairs. So it is simply a matter of building your internal buffers and such
remembering to start them off with the G1 symbol first before sending to the viterbi decoder.
As you might start to guess, it has all been designed to work easily - the 36-way interleaving means exactly 2 times
around for the switching per 72 bit frame of data. And each frame starts with the G!/G2 symbol pairings aligned
correctly at the beginning of each frame.
BACK