5G Wireless Standard Design Notes

Module 6 :  

Proportional Fair scheduling algorithms used in Base stations (Lower data rate users are artificially boosted to ensure they get access to the resources)

Module 7: 5G physical downlink shared channel (PDSCH) transmit chain– CRC generation

• Min Transport Block size 1RB = 24 bits MCS0 max Transport size = 319784, 275RB MCS27 
• A 24 bit CRC can detect upto and including 24 bit errors
• An Example of 2 bit CRC calculation given
• Physical layer starts with CRC and ends with CRC (38.212, section 5.1)

Module 8: 5G PDSCH transmit chain – code block segmentation – part I

Segment Size = 8448, + 24-bit CRC

   

LDPC Decoder/Encoder needs a higher number of bits for better coding gain i.,e it can perform better even with a low SNR. Larger the codeblock size, better will be the performance of the decoder. In the above pic example the SNR needed for 8448 bits is lower compared to SNR needed to decode 2848 bits, i.e., we are penalizing the guy with lower codeblock. BLER is limited by the smallest transport size.

On the other hand, a high number of bits (Segment size) can lead to a higher decoding implementation complexity.

Module 9: 5G PDSCH transmit chain – LDPC coding

In the u matrix, 2 means the colums of I need to be shifted by 2, 3 means colums of I need to be shifted by 3 etc., This is done so that we dont need to store entire H matrix (savings in space in systems). Instead given a U and a I and size, we can generate a H matrix.

5G PDSCH supports 2 base graphs

1.     Base Graph 1 optimized for large information block size and high code rates
1. max code rate 8/9 min code rate 1/3
2. Base Graph 2 optimized for small information block size and low code rates.
1. Lowest code rate is 1/5

Base Graph 2 has lower decoding complexity.

Kb - The number of columns in Systematic matrix is fixed. implementation constraint. therfore the number of rows = Information bits (i.e., 8448) /22 = 384. so summarize we ll have a 384 X 22 matrix.

Module 10: 5G PDSCH transmit chain – code block segmentation – part II