bravoegg
Member level 2
OFDM symbols use cyclic prefix and CP could be used to calculate the start of OFDM symbols.
It's usually by correlation. I've done the simulations, but the resultant estimated OFDM starting point is not stable. They vary within a range of [-20,20].
I know these difference could be eliminated latter. But when I'm trying to simulate the integer frequency offset. Theses variations become troublesome.
If all coarse symbol starting point estimates are at the same point, say -1, that is 1 sample before the actual OFDM symbol. This would be just fine. The integer frequency offset could be calculated perfectly, at a bad SNR as low as 2db.
However, when each OFDM symbol has a different symbol starting estimate point, the integer frequency offset is no longer perfectly calculated. They fail from time to time.
I've been stuck here for weeks. Could someone show me how actual coarse symbol synchronization is done in practice?
Thanks in advance.
It's usually by correlation. I've done the simulations, but the resultant estimated OFDM starting point is not stable. They vary within a range of [-20,20].
I know these difference could be eliminated latter. But when I'm trying to simulate the integer frequency offset. Theses variations become troublesome.
If all coarse symbol starting point estimates are at the same point, say -1, that is 1 sample before the actual OFDM symbol. This would be just fine. The integer frequency offset could be calculated perfectly, at a bad SNR as low as 2db.
However, when each OFDM symbol has a different symbol starting estimate point, the integer frequency offset is no longer perfectly calculated. They fail from time to time.
I've been stuck here for weeks. Could someone show me how actual coarse symbol synchronization is done in practice?
Thanks in advance.
