In all documents I saw that theoretical rate for these conditions is 64.8 Mbps. So the difference between 75.6 Mbps and 64.8 Mbps is control channels etc.
Next question - How to calculate "real" theoretical max throughput?
Each PRB looks like the picture below. As we said the PRB consists of 12 subcarriers and the slot (0.5ms) carries 7 symbols. This is represented as a two-dimensional matrix.
Each subcarrier/symbol combination is referred to as a RB (Resource Element).
In total we have 12 x 7 = 84 RE per PRB.
Control channels will use 12 RE per PRB (6 for the PDCCH and 6 for the Reference Signals)
So now you are left with 72 RE per PRB.
72 x 50 (total PRB per 10MHz) x 2 (to give value in ms) = 7200 RE per ms
Each RE can carry 6 bits (64 QAM) so 43200 bits per ms or 43.2Mbps
In addition, per 10ms frame additional Physical Channel Overheads are also present, i.e. the PSS, SSS and PBCH. This would bring the rate slightly down 64.5516Mbps (assuming a 3/4 rate coding).
Added: Note that this assumes a Normal Cyclic prefix. If the extended CP is used the rate for the same 3/4 mimo 2x2 would be 53.7516Mbps.
To answer pablo123 question - link adaption is used. However the system relies more on HARQ mechanisms with RRC configuring the number of HARQ retries (note that the uplink and downlink different HARQ methods).
If these fail RLC (in AM mode) includes re-segmentation of existing RLC AM PDUs - which can then be sent with link adapted Modumation and coding....with HARQ...VERY Flexible!
Note that the downlink and uplink are slightly different in terms of effective coding rates (based on TB(Transport Block Size)) the effective coding rate in the downlink in Release 8 is limited to 0.9 (Not 1).
Cheers
Redrooster
Last edited by redrooster; 2010-08-27 at 12:41 AMReason: forgot to add about Extended CP
Bookmarks