silas/imdd_silas
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

MPI Simulations and stuff

Browse Source
This commit is contained in:
Silas Oettinghaus
2024-08-14 09:36:51 +02:00
parent 1eeb970d8f
commit 34f9149346
61 changed files with 4295 additions and 429 deletions
Download Patch File Download Diff File Expand all files Collapse all files

192
Classes/04_DSP/EQ_silas.m
View File

@@ -170,23 +170,23 @@ classdef EQ_silas < handle
for n = obj.sps*obj.delay+1:obj.sps:obj.sps*obj.trainlength
m = m+1;
dc_cnt = dc_cnt+1;
%get Sigal input vectors with correct length for VNLE
x_in_block = obj.x_in(obj.Ne(1)+n+(obj.sps-1):-1:n+obj.sps).';
x_in_vnle_format = obj.calcVNLENonlinVecs(x_in_block,obj.Ie2,obj.Ie3,obj.Ne,obj.x_norm);
%get Reference input vectors with correct length for VNLE
d_block = obj.d(obj.Nb(1)-obj.delay+m-2:-1:m-obj.delay-1).';
d_vnle_format = obj.calcVNLENonlinVecs(d_block,obj.Ib2,obj.Ib3,obj.Nb,obj.d_norm);
obj.e_ffe = obj.e.' * x_in_vnle_format;
obj.e_dfe = obj.b.' * d_vnle_format;
% Calculate the Error
obj.error = obj.e_dc + obj.e_ffe - obj.e_dfe - obj.d(obj.Nb(1)-1+m-obj.delay);
if obj.mu_ffe_train ~= 0
%update FFE coefficients with LMS
obj.e = obj.e - obj.error*conj(x_in_vnle_format)*obj.mu_ffe_train;
@@ -194,18 +194,17 @@ classdef EQ_silas < handle
%update FFE coefficients with NLMS
obj.e = obj.e - obj.error*x_in_vnle_format/(x_in_vnle_format.'*x_in_vnle_format);
end
%update DFE coefficients with LMS
obj.b = obj.b + obj.mu_dfe_train*obj.error*d_vnle_format;
%update DC error
dc_block(dc_cnt) = obj.error .* obj.mu_dc_train;
if dc_cnt == obj.eq_parallelization_blocklength
obj.e_dc = obj.e_dc - mean(dc_block(dc_cnt));
dc_cnt = 0;
end
end
@@ -227,42 +226,29 @@ classdef EQ_silas < handle
dc_cnt = 0;
mu_mat = diag([ones(1,obj.Ce(1))*obj.mu_ffe_dd(1)... %1st order ffe
ones(1,obj.Ce(2))*obj.mu_ffe_dd(2)... %2nd order ffe
ones(1,obj.Ce(3))*obj.mu_ffe_dd(3)... %3rd order ffe
ones(1,sum(obj.Cb))*obj.mu_dfe_dd]); %all order dfe
mu_ffe = [ones(1,obj.Ce(1))*obj.mu_ffe_dd(1)... %1st order ffe
ones(1,obj.Ce(2))*obj.mu_ffe_dd(2)... %2nd order ffe
ones(1,obj.Ce(3))*obj.mu_ffe_dd(3)];
mu_dfe = ones(1,sum(obj.Cb))*obj.mu_dfe_dd;
ones(1,obj.Ce(2))*obj.mu_ffe_dd(2)... %2nd order ffe
ones(1,obj.Ce(3))*obj.mu_ffe_dd(3)... %3rd order ffe
ones(1,sum(obj.Cb))*obj.mu_dfe_dd]); %all order dfe
y = zeros(1,floor(obj.x_length/obj.sps));
d_feedback = zeros(obj.Cb(1),1);
d_vnle = obj.calcVNLENonlinVecs(d_feedback,obj.Ib2,obj.Ib3,obj.Nb,obj.d_norm);
d_hat = zeros(obj.x_length,1);
lvl_err = NaN(obj.x_length,numel(obj.d_constellation));
lvl_err_mov = NaN(100,numel(obj.d_constellation));
d_hat = NaN(length(obj.d),numel(obj.d_constellation));
lvl_err_1 = NaN(length(obj.d),numel(obj.d_constellation));
lvl_err_2 = NaN(length(obj.d),numel(obj.d_constellation));
subtracted_error =NaN(length(obj.d),numel(obj.d_constellation));
y_1= NaN(length(obj.d),numel(obj.d_constellation));
y_2= NaN(length(obj.d),numel(obj.d_constellation));
lvl_err_mov = NaN(obj.eq_avg_blocklength,numel(obj.d_constellation));
m_reg = 0;
if obj.eq_avg_blocklength > 0
averaging_window = zeros(obj.eq_avg_blocklength,1);
end
for k = 1:obj.sps:obj.x_length
dc_cnt = dc_cnt+1;
m=m+1;
%get Sigal input vectors with correct length for VNLE
x = obj.x_in(obj.Ne(1)+k-1:-1:k).';
%
if obj.eq_avg_blocklength > 0 %% Das läuft gut mit 400er Fenster!!
averaging_window = circshift(averaging_window,obj.sps);
averaging_window(1:obj.sps,1) = x(1:obj.sps);
avg_(k) = mean(averaging_window);
x = x-avg_(k);
end
%bring this signal to "special" VNLE format
x_vnle = obj.calcVNLENonlinVecs(x,obj.Ie2,obj.Ie3,obj.Ne,obj.x_norm);
@@ -270,61 +256,42 @@ classdef EQ_silas < handle
x_d = [x_vnle;-d_vnle];
%Apply filter
if obj.mu_dc_dd > 0
y(m) = obj.e_dc(end) + x_d.'* coeff;
else
y(m) = x_d.'* coeff;
end
y(m) = x_d.'* coeff;
%Decision 1
[~,symbol_idx] = min(abs(y(m) - obj.d_constellation)); % decision for closest constellation point
d_hat(k) = obj.d_constellation(symbol_idx);
%Error between FFE & DFE filtered signal and Decision
obj.error(k) = y(m) - d_hat(k);
%
lvl_err(k,symbol_idx) = obj.error(k);
d_hat(m,symbol_idx) = obj.d_constellation(symbol_idx);
lvl_err_mov(:,symbol_idx) = circshift(lvl_err_mov(:,symbol_idx),1);
lvl_err_mov(1,symbol_idx) = obj.error(k);
y_1(m,symbol_idx) = y(m); % after 1st iteration
%Decision 2
y(m) = y(m)-mean(lvl_err_mov(:,symbol_idx),'omitnan');
[~,symbol_idx] = min(abs(y(m) - obj.d_constellation)); % decision for closest constellation point
d_hat(k) = obj.d_constellation(symbol_idx);
%1st Error between FFE & DFE filtered signal and Decision
obj.error(m) = y(m) - d_hat(m,symbol_idx);
% lvl_err_1(m,symbol_idx) = y(m) - obj.d(m+1);
%
% %write current error to buffer
% lvl_err_mov(:,symbol_idx) = circshift(lvl_err_mov(:,symbol_idx),1);
% lvl_err_mov(1,symbol_idx) = obj.error(m);
%
% %Subtract a weighted error from y -> then Decision 2
% err = mean(lvl_err_mov(:,symbol_idx),'omitnan');
%
% y(m) = y(m)-(obj.mu_dc_dd(symbol_idx)*err);
%
% subtracted_error(m,symbol_idx) = obj.mu_dc_dd(symbol_idx)*mean(lvl_err_mov(:,symbol_idx),'omitnan');
%
% y_2(m,symbol_idx) = y(m);
%
% [~,symbol_idx] = min(abs(y(m) - obj.d_constellation)); % decision 2 for closest constellation point
%
% d_hat(m,symbol_idx) = obj.d_constellation(symbol_idx);
%
% obj.error(m) = y(m) - d_hat(m,symbol_idx);
%
% lvl_err_2(m,symbol_idx) = y(m) - obj.d(m+1);
%Update FFE and DFE coefficients
coeff = coeff - (mu_mat * (obj.error(k) * conj(x_d)));
%Update DC error
dc_block(dc_cnt) = obj.error(k) ;
if dc_cnt == obj.eq_parallelization_blocklength
if obj.eq_updatelatency > 1
obj.e_dc = circshift(obj.e_dc,1);
% m_reg(end+1) = ((1:obj.eq_parallelization_blocklength)' \ (cumsum(dc_block)));
%
% obj.e_dc(1) = obj.e_dc(2) - sign(m_reg(end)) .* (sum(dc_block).* m_reg(end) .* obj.mu_dc_dd);
obj.e_dc(1) = obj.e_dc(2) - sum(dc_block) .* obj.mu_dc_dd;
else
%m_reg(end+1) = ((1:obj.eq_parallelization_blocklength)' \ (cumsum(dc_block)));
% obj.e_dc = obj.e_dc - sign(m_reg(end)) .* (sum(dc_block).* m_reg(end) .* obj.mu_dc_dd);
obj.e_dc = obj.e_dc - sum(dc_block) .* obj.mu_dc_dd;
% obj.e_dc = obj.e_dc - obj.mu_dc_dd * obj.error(k); %newapril
end
dc_cnt = 0;
end
coeff = coeff - (mu_mat * (obj.error(m) * conj(x_d)));
% Append new decision to decision feedback
if obj.Nb(1) > 0
@@ -332,43 +299,52 @@ classdef EQ_silas < handle
%shift up one index
d_feedback(2:end) = d_feedback(1:end-1);
%replace 1st index with current estimation
d_feedback(1) = d_hat(k);
d_feedback(1) = d_hat(m,symbol_idx);
%build memorylike VNLE version
d_vnle = obj.calcVNLENonlinVecs(d_feedback,obj.Ib2,obj.Ib3,obj.Nb,obj.d_norm);
end
end
end
%%
%
% b = movmean(lvl_err,[500 500],1,"omitnan");
%
% figure(11)
% for i = 1:4
% hold on
% stem(lvl_err(:,i))
% end
%%
obj.y_out = (circshift( y.' ,-(obj.delay))).';
obj.d_out = d_hat(1:2:end);
% err = obj.error(1:2:end);
% res = NaN(8,length(err));
% for lvl = 1:8
% a = find(obj.d_out==obj.d_constellation(lvl));
% res(lvl,a) = err(a);
% end
% mean(res,2,"omitnan");
%
% figure(12)
% scatter(1:length(obj.y_out),obj.y_out,1,'.')
% hold on
% scatter(1:length(obj.d_out),obj.d_out,1,'.')
% evm1 = mean(lvl_err_1,'omitnan');
%
% evm2 = mean(lvl_err_2,'omitnan');
%
% figure(112)
% stem(evm1,'LineStyle','--','Marker','square','LineWidth',1);
% hold on;
% stem(evm2,'LineStyle',':','Marker','v','LineWidth',1);
% figure(14)
% scatter(1:length(lvl_err_1),subtracted_error,1,'.')
%
% lvl_err___ = lvl_err_true(~isnan(lvl_err_true));
% %lvl_err___ = lvl_err___-mean(lvl_err___);
% coeffs = arburg(lvl_err___,1000);
% fs_in = 92e9;
% [h,w] = freqz(1,coeffs,length(lvl_err___),"whole",fs_in);
% h = fftshift(h./max(abs(h)));
% freq_vec = linspace(-fs_in/2,fs_in/2,length(h));
% figure(111)
% hold on
% plot(freq_vec.*1e-9,20*log10(h),'DisplayName','burg');
%
% spectrum_plot(y,92e9);
%
% d = 2^nextpow2(length(y)/16);
%
% figure(1111)
% hold on
% pwelch(y,hamming(d),d/2,d,92e9,"centered","power");
end