In 4G and 5G bits after channel encoding are scrambled with a pseudo random sequence of bits. The output of the channel encoder can theoretically be same with other interfering signal which may lead to failure of decoding at the receiver end. This scrambling module adds extra randomness to the encoded bits which provides coding gain by supressing the interference.
This scrambling is applied to all the channels in 5G. The pseudo random sequence is different and unique for each channel and each cell.
In short, pseudo random sequence is generated by doing XOR of two m-sequences. 1st m-sequence is initialized with 1 where 2nd m-sequence is initialized with a seed value (known as Cinit).
Here are the different seed values (Cinit) for each channel in 5G:
PBCH: f( cellID ) — where f(x): function of x
PDSCH: f( 𝑛RNTI, 𝑞, 𝑛ID )
PDCCH: f( 𝑛RNTI, 𝑛ID )
PUSCH: f( 𝑛RNTI, 𝑛ID )
PUCCH: f( 𝑛RNTI, 𝑛ID ) (For PUCCH format 2, 3 and 4)
CellID: physical cell id which is different for each cell
There are different RNTIs in 5G. In short it is a unique identity provided to the UE by the network. If PDSCH contains broadcast information then 𝑛RNTI will be message specific RNTI (example: for SIB1 it will be SI-RNTI) and if PDSCH contains UE specific data then it will be unique to that UE.
q stands for codeword index. Since there can be max. two codewords in PDSCH q takes values 0 or 1.
𝑛ID is either the cell ID or a unique ID
Since PBCH is the first broadcast message decoded by the UE and it is not unique to UEs, it unique to cells it’s scrambling seed value is only function of Cell ID.
For rest of the channels seed value is unique to the UEs, codewords (for PDSCH) and for cell.
𝑛RNTI, 𝑞 and 𝑛ID make sure Cinit is different for each codeword, each UE and for each network.
Let’s say there are two cell towers at different distances to a UE. Both the cells are synchronized in time and uses the same band. Both the cells are transmitting different SIB1 (System Information Block 1) but both happen to transmit same DCI (DCI 1_0 with SI-RNTI) to decode their SIB1. (Above situation is not highly unlikely)
If we don’t scramble the encoded bits, UE might decode SIB1 of another cell it is not connected to or it might fail to decode any of these.
You must be logged in to post a comment.