当前位置：网站首页>UCI and data multiplexing are transmitted on Pusch (Part VI) -- LDPC coding
UCI and data multiplexing are transmitted on Pusch (Part VI) -- LDPC coding
2022-07-03 09:40:00 【Communication pawn】
ldpc The coding is for the coding of uplink and downlink data , It's also 5G The most important kind of coding .
1 TB Block add CRC
This is and UCI POLAR One difference ,UCI Yes for each. cb Block add crc.
% Get transport block size after CRC attachment according to 38.212
% 6.2.1 and 7.2.1, and assign CRC polynomial to CRC field of output
% structure info
if A > 3824
L = 24;
info.CRC = '24A';
else
L = 16;
info.CRC = '16';
end
** This will be true of TB Block on crc The addition of , Add crc The process has been in the last article
《UCI And data reuse in pusch Up transfer （ The fifth part ）---polar code 》 Shared ,
There is no retelling here .**
2 graph choice
% LDPC base graph selection
if A <= 292 || (A <= 3824 && R <= 0.67) || R <= 0.25
bgn = 2;
else
bgn = 1;
end
3 cb Block segmentation and cb Block add crc
Add crc The process has been in the last article
《UCI And data reuse in pusch Up transfer （ The fifth part ）—polar code 》 Shared ,
There is no retelling here .
% Get the maximum code block size
if bgn == 1
Kcb = 8448;
else
Kcb = 3840;
end
% Get number of code blocks and length of CB-CRC coded block
if B <= Kcb
L = 0;
C = 1;
Bd = B;
else
L = 24; % Length of the CRC bits attached to each code block
C = ceil(B/(Kcb-L));
Bd = B+C*L;
end
% Obtain the number of bits per code block (excluding CB-CRC bits)
cbz = ceil(B/C);
% Get number of bits in each code block (excluding filler bits)
Kd = ceil(Bd/C);
% Find the minimum value of Z in all sets of lifting sizes in 38.212
% Table 5.3.2-1, denoted as Zc, such that Kb*Zc>=Kd
if bgn == 1
Kb = 22;
else
if B > 640
Kb = 10;
elseif B > 560
Kb = 9;
elseif B > 192
Kb = 8;
else
Kb = 6;
end
end
Zlist = [2:16 18:2:32 36:4:64 72:8:128 144:16:256 288:32:384];
Zc = min(Zlist(Kb*Zlist >= Kd));
% Get number of bits (including <NULL> filler bits) to be input to the LDPC
% encoder
if bgn == 1
K = 22*Zc; % Why must this be Zc Multiple , This is a problem ？
else
K = 10*Zc;
end
4 UL_SCH Channel code of （ This is the most difficult part of coding, which needs to be further clarified ）
% Get lifting set number
ZSets = {
[2 4 8 16 32 64 128 256],... % Set 1
[3 6 12 24 48 96 192 384],... % Set 2
[5 10 20 40 80 160 320],... % Set 3
[7 14 28 56 112 224],... % Set 4
[9 18 36 72 144 288],... % Set 5
[11 22 44 88 176 352],... % Set 6
[13 26 52 104 208],... % Set 7
[15 30 60 120 240]}; % Set 8
for setIdx = 1:8 % LDPC lifting size set index
if any(Zc==ZSets{
setIdx})
break;
end
end
% Pre-stored H types for each BGN and setIdx pair
% How did this come from? I don't know , Need to take a further look
Htype = {
3, 3, 3, 3, 3, 3, 2, 3;
4, 4, 4, 1, 4, 4, 4, 1};
% Get the matrix with base graph number 'bgn' and set number 'setIdx'.
% The element of matrix V in the following is H_BG(i,j)*V(i,j), where
% H_BG(i,j) and V(i,j) are defined in TS 38.212 5.3.2; if V(i,j) is not
% defined in Table 5.3.2-2 or Table 5.3.2-3, the elements are -1.
% % This is in the agreement v_ij
switch bgn
case 1
switch setIdx
case 1
V = bgs.BG1S1;
case 2
V = bgs.BG1S2;
case 3
V = bgs.BG1S3;
case 4
V = bgs.BG1S4;
case 5
V = bgs.BG1S5;
case 6
V = bgs.BG1S6;
case 7
V = bgs.BG1S7;
otherwise % 8
V = bgs.BG1S8;
end
Nplus2Zc = Zc*(66+2);
otherwise % bgn = 2
switch setIdx
case 1
V = bgs.BG2S1;
case 2
V = bgs.BG2S2;
case 3
V = bgs.BG2S3;
case 4
V = bgs.BG2S4;
case 5
V = bgs.BG2S5;
case 6
V = bgs.BG2S6;
case 7
V = bgs.BG2S7;
otherwise % 8
V = bgs.BG2S8;
end
Nplus2Zc = Zc*(50+2);% Why ask this and ？
end
P = nr5g.internal.ldpc.calcShiftValues(V,Zc);% according to V and Zc obtain P P=mod(V,Zc)
% The element of matrix P in the following is P(i,j) in TS 38.212 5.3.2
% when V(i,j) are defined in Table 5.3.2-2 or Table 5.3.2-3. If not
% defined, the elements are -1.
P = zeros(size(V));
for i = 1:size(V,1)
for j = 1:size(V,2)
if V(i,j) == -1
P(i,j) = -1;
else
P(i,j) = mod(V(i,j),Z);
end
end
end
P1 = P(:,1:(size(P,2) - size(P,1)));% Take it P In front of （52-42） perhaps （67-45） That's ok , Why take this ？
C = size(infoBits,2); % Get layers
codewords = zeros(Nplus2Zc,C);
% Get shift values matrix
P = nr5g.internal.ldpc.calcShiftValues(V,Zc);
P1 = P(:,1:(size(P,2) - size(P,1)));
C = size(infoBits,2);
codewords = zeros(Nplus2Zc,C);
for r = 1:C
% Per code-block processing only
infoVec = reshape(infoBits(:,r),Zc,[]);% The information bit（ In the agreement C_k） become Zc That's ok ,K/Zc Column .
d = blockMultiply(P1,infoVec);
d0 = d(:,1:4);
% Solve 4 equations for Htype * m = d0
switch Htype{
bgn,setIdx}
case 1
% H1 = [ 1 0 -1 -1;
% -1 0 0 -1;
% 0 -1 0 0;
% 1 -1 -1 0 ];
m1 = sum(d0, 2);
m2 = d0(:,1) + m1([2:end 1]);
m3 = d0(:,2) + m2;
m4 = d0(:,3) + m1 + m3;
case 2
% H2 = [ 0 0 -1 -1;
% 105 0 0 -1;
% -1 -1 0 0;
% 0 -1 -1 0 ];
m1 = sum(d0, 2);
shift = mod(105, Zc);
if shift > 0
m1 = m1([(end-shift+1):end 1:(end-shift)]);
end
m2 = d0(:,1) + m1;
m4 = d0(:,4) + m1;
m3 = d0(:,3) + m4;
case 3
% H3 = [ 1 0 -1 -1;
% 0 0 0 -1;
% -1 -1 0 0;
% 1 -1 -1 0 ];
m1 = sum(d0, 2);
m2 = d0(:,1) + m1([2:end 1]);
m3 = d0(:,2) + m1 + m2;
m4 = d0(:,3) + m3;
otherwise % == 4
% H4 = [ 0 0 -1 -1;
% -1 0 0 -1;
% 1 -1 0 0;
% 0 -1 -1 0 ];
m1 = sum(d0, 2);
m1 = m1([end 1:(end-1)]);
m2 = d0(:,1) + m1;
m3 = d0(:,2) + m2;
m4 = d0(:,4) + m1;
end
% Get other parity bits
P3 = P(5:end,size(P1,2)+(1:4));
p = blockMultiply(P3, [m1 m2 m3 m4]) + d(:,5:end);
% Form codeword and assign to output
codewords(:,r) = [infoBits(:,r); mod([m1; m2; m3; m4; p(:)],2)];
end
end
5 Rate matching
% Get code block soft buffer size
if ~isempty(Nref)
fcnName = 'nrRateMatchLDPC';
validateattributes(Nref, {
'numeric'}, ...
{
'scalar','integer','positive'},fcnName,'NREF');
Ncb = min(N,Nref);
else % No limit on buffer size
Ncb = N;
end
% Get starting position in circular buffer
if bgn == 1
if rv == 0
k0 = 0;
elseif rv == 1
k0 = floor(17*Ncb/N)*Zc;
elseif rv == 2
k0 = floor(33*Ncb/N)*Zc;
else % rv is equal to 3
k0 = floor(56*Ncb/N)*Zc;
end
else
if rv == 0
k0 = 0;
elseif rv == 1
k0 = floor(13*Ncb/N)*Zc;
elseif rv == 2
k0 = floor(25*Ncb/N)*Zc;
else % rv is equal to 3
k0 = floor(43*Ncb/N)*Zc;
end
end
% Get rate matching output for all scheduled code blocks and perform
% code block concatenation according to Section 5.4.2 and 5.5
out = [];
for r = 0:C-1
if r <= C-mod(outlen/(nlayers*Qm),C)-1
E = nlayers*Qm*floor(outlen/(nlayers*Qm*C));
else
E = nlayers*Qm*ceil(outlen/(nlayers*Qm*C));
end
out = [out; cbsRateMatch(in(:,r+1),E,k0,Ncb,Qm)]; %#ok<AGROW>
end
end
function e = cbsRateMatch(d,E,k0,Ncb,Qm)
% Rate match a single code block segment as per TS 38.212 Section 5.4.2
% Bit selection, Section 5.4.2.1
k = 0;
j = 0;
e = zeros(E,1,class(d));
while k < E
if d(mod(k0+j,Ncb)+1) ~= -1 % Filler bits
e(k+1) = d(mod(k0+j,Ncb)+1);
k = k+1;
end
j = j+1;
end
% Bit interleaving, Section 5.4.2.2
e = reshape(e,E/Qm,Qm);
e = e.';
e = e(:);
end
边栏推荐
- How MySQL modifies null to not null
- Error output redirection
- Find all possible recipes from given supplies
- 2021-09-26
- numpy. Reshape() and resize() functions
- Arduino handles JSON data, arduinojson assistant
- MySQL Data Definition Language DDL common commands
- PolyWorks script development learning notes (II) -treeview basic operations
- Getting started with shell programming
- Starting from 0, use pnpm to build a demo managed by monorepo
猜你喜欢
![[male nanny style] teach you to open the first wechat applet](/img/a1/a571609ee846adf75506a88a629906.png)
[male nanny style] teach you to open the first wechat applet
What do software test engineers do? Pass the technology to test whether there are loopholes in the software program
The rise and fall of mobile phones in my perspective these 10 years
UCI and data multiplexing are transmitted on Pusch (Part 4) --small block lengths
Please tell me how to set vscode
【男保姆式】教你打开第一个微信小程序
Win10 install elk
Difference of EOF
Leetcode daily question (2090. K radius subarray averages)
LeetCode每日一题(2212. Maximum Points in an Archery Competition)
随机推荐
LeetCode每日一题(985. Sum of Even Numbers After Queries)
Filter comments to filter out uncommented and default values
Make the most basic root file system of Jetson nano and mount NFS file system on the server
Nr-prach:prach format and time-frequency domain
[male nanny style] teach you to open the first wechat applet
Leetcode daily question (745. prefix and suffix search)
Solve editor MD uploads pictures and cannot get the picture address
Idea uses the MVN command to package and report an error, which is not available
Leetcode daily question (516. long palindromic subsequence)
Nodemcu-esp8266 development board to build Arduino ide development environment
Electronic product design, MCU development, circuit cloning
Arduino handles JSON data, arduinojson assistant
Flask+supervisor installation realizes background process resident
Leetcode daily question (985. sum of even numbers after queries)
Learn the contents of 5g toolbox supporting NR through the NR resources provided by MATLAB
[graduation successful] [1] - tour [Student Management Information System]
顺利毕业[2]-学生健康管理系统 功能开发中。。。
Flink CDC practice (including practical steps and screenshots)
全球KYC服务商ADVANCE.AI 活体检测产品通过ISO国际安全认证 产品能力再上一新台阶
What do software test engineers do? Pass the technology to test whether there are loopholes in the software program