Wednesday, May 23, 2012 6:47 AM
Hi, I have some questions about Sora.
First, with 802.11a protocol, each complex16 is truncated to a complex8, but I find that RAB uses a 14-bit DAC. If considering PAPR which can be 6dB or so for OFDM symbols, we may use 13 bits for each complex. Why does Sora use a complex8?
Second, I find Sora SDK uses 10720 to denote +1 when mapping, but when demaaping, the demapping tables seems to denote +1 by 7. Why does Sora use 7 to denote +1 when demapping?
- Edited by ychm Wednesday, May 23, 2012 6:47 AM
Saturday, May 26, 2012 4:06 AMModerator
For your first question:
PAPR is a valid consideration. In Sora, using complex8 is a history problem rather than a design choice. However, according to our experience, 8-bit IQ (only on TX path) is not a bottleneck in current Sora realization, because (1) 8-bit already provides 6x8=48dB dynamic range (2) Although DAC is 14-bit, the valid dynamic range of radio device will not be as high as 6x14dB, because board implementation usually introduces noise and the lower bits are useless (3) In extreme case, i.e. PAPR is very high, 8-bit IQ could be a problem. But the case of high PAPR is rare.
Complex8 will be replaced by Complex16 in later Sora version, e.g. MIMO.
- Marked As Answer by Jiansong zhangMicrosoft Employee, Moderator Saturday, June 02, 2012 4:09 AM