I have reviewed and executed the provided code in MATLAB R2023b. The results showed a Bit Error Rate (BER) within the range of 
when AWGN noise with a variance of 3 was applied, and zero BER when no AWGN noise was used. You can find the implementation below.
when AWGN noise with a variance of 3 was applied, and zero BER when no AWGN noise was used. You can find the implementation below.clear;
K = 128;
E = 2*K;
mbits = randi([0,1],K, 1);
crc = lteCRCEncode(mbits,'24A');
cbs = lteCodeBlockSegment(crc); % This will provide a length 4096 bits that is input to the Turbo encoder
cd = lteTurboEncode(cbs);
cdrm = lteRateMatchTurbo(cd, E, 0);
%cdrm(cdrm == 0) = -1; % Make them as LLRs
txSymbols = lteSymbolModulate(cdrm,'QPSK');
awgnchan = comm.AWGNChannel('NoiseMethod','Variance','Variance',3);
rxSymbols = txSymbols;% awgnchan(txSymbols);
softBits= lteSymbolDemodulate(rxSymbols,'QPSK','Soft');
cdrx = lteRateRecoverTurbo(softBits, K, 0);
%
mhat = lteTurboDecode(cdrx); % So after the Turbo decoding, it will have 4096 bits rather than 4032 bits, due to addition of filler bits and CRC
cbshat = lteCodeBlockDesegment(mhat, K+24);
crchat = lteCRCDecode(cbshat,'24A'); % This will return the same as message length
% Check the decoded bits with message bits
[~,BER] = biterr( double(crchat),mbits);
fprintf("Bit Error Rate without AWGN noise: %d ",BER)
Additionally, I found that the lteSymbolModulate function supports BPSK modulation.
txSymbols = lteSymbolModulate(cdrm,'BPSK'); % For Modulation
softBits = lteSymbolDemodulate(rxSymbols,'BPSK','Soft'); % For Demodulation
Here is the relevant documentation:
If there are any other requirements or modifications needed, let me know.
