nrDCIDecode

Decode downlink control information (DCI)

Syntax

dcibits = nrDCIDecode(softbits,K,L)

[dcibits,mask] = nrDCIDecode(softbits,K,L)

[dcibits,mask] = nrDCIDecode(softbits,K,L,rnti)

Description

dcibits = nrDCIDecode(softbits,K,L) decodes the input softbits and returns the decoded DCI bits of length K. The function implements the inverse of the features specified in TS 38.212 Sections 7.3.4, 7.3.3, and 7.3.2 [1], such as rate recovery, polar decoding, and cyclic redundancy check (CRC) decoding. L specifies the list length used for polar decoding.

[dcibits,mask] = nrDCIDecode(softbits,K,L) also looks for a cyclic redundancy check (CRC) error in the DCI decoding. If mask is not equal to 0, either an error has occurred or the input CRC has been masked. When there are no CRC errors, mask is the actual value used for masking the CRC bits.

[dcibits,mask] = nrDCIDecode(softbits,K,L,rnti) specifies a radio network temporary identifier (RNTI). You can use this syntax when the value of rnti masks the CRC parity bits at the transmit end. When you specify rnti and there are no CRC errors, mask equals to 0.

Examples

Decode DCI Codeword

Create a random sequence of binary values corresponding to a DCI message of 32 bits. Encode the message based on the specified RNTI and rate-matched DCI codeword length. The RNTI masks the CRC parity bits.

K = 32; 
rnti = 100; 
E = 240; 
dciBits = randi([0 1],K,1);
dcicw = nrDCIEncode(dciBits,rnti,E);

Decode the soft bits representing the DCI codeword dcicw by specifying the RNTI used for the CRC masking. Set the length of the polar decoding list to 8.

L = 8;
[recBits,mask] = nrDCIDecode(1-2*dcicw,K,L,rnti)

recBits = 32x1 int8 column vector

   1
   1
   0
   1
   1
   0
   0
   1
   1
   1
      ⋮

mask = uint32

0

Verify that the transmitted and received message bits are identical.

isequal(recBits,dciBits)

ans = logical
   1

Verify that the decoding is without error. As the decoding specified the RNTI used for masking, a mask value of 0 indicates no error.

mask

mask = uint32

0

Input Arguments

softbits — Coded block of soft bits

column vector of real numbers

Coded block of soft bits, specified as a column vector of real numbers.

Data Types: double | single

K — Length of decoded output in bits

integer

Length of decoded output in bits, specified as an integer from 12 to 140.

Data Types: double

L — Length of polar decoding list

power of two

Length of polar decoding list, specified as a power of two.

Data Types: double

rnti — RNTI of UE

integer from 0 to 65,535

RNTI of the UE, specified as an integer from 0 to 65,535.

Data Types: double

Output Arguments

dcibits — Decoded DCI message bits

K-by-1 column vector of binary values

Decoded DCI message bits, returned as a K-by-1 column vector of binary values. The message bits were transmitted on a single physical downlink control channel (PDCCH).

Data Types: int8

mask — Result of CRC decoding

nonnegative integer

Result of CRC decoding, returned as a nonnegative integer less than or equal to 2¹⁶–1. If mask is not equal to 0, either an error has occurred or the CRC has been masked. When there are no errors, mask is the actual value used for masking the CRC bits.

Data Types: uint32

References

[1] 3GPP TS 38.212. “NR; Multiplexing and channel coding.” 3rd Generation Partnership Project; Technical Specification Group Radio Access Network.

Extended Capabilities

C/C++ Code Generation

Generate C and C++ code using MATLAB® Coder™.

Usage notes and limitations:

The input argument L must be less than or equal to 64.

Version History

Introduced in R2018b

R2023b: C/C++ code generation updates

For C/C++ code generation, the limitation that the input argument L must be a compile-time constant has been removed, however, L must be less than or equal to 64.

R2020a: Polar decoding metric update

In releases R2019b and before, polar decoding uses the exact form of the expression $$\log (1+e^x)$$ for internal metric evaluation. Starting in release R2020a, because the exact form leads to numerical instability for high SNR ranges, polar decoding approximates $$\log (1+e^x)$$ as 0 for $$x<0$$ and as $$x$$ for $$x\ge 0$$. This approximation affects the results of the nrDCIDecode function, resulting in a marginal degradation of the BLER performance in a link-level simulation.