You are on page 1of 3

Iterative Methods (WS2014/2015) Matlab Project 3

Exercise III: Extended Mapping for Bit-Interleaved


Coded Modulation

1 Introduction
In project 3 we simulate a bit-interleaved coded modulation transmission scheme with
extended mapping applied to the transmit alphabet. This includes:

• Designing the bit-to-symbol mapper that accounts for multiple bit labels for each
symbol.

• Designing the soft-input soft-output demapper that calculates the log-likelihood


ratio (LLR) of the coded bits.

The aim of this project is to introduce an advanced transmission scheme that requires
iterative demapping and decoding.

1.1 Transmission Model

q c0 c x
COD Π MAP

q̂ LDEM Ld
DEC Π−1 DEMAP y


Π
Lcc LDEC

Figure 1: System model for BICM-ID over AWGN channel.

Bit-interleaved coded modulation (BICM) [1], is a transmission scheme where by bit-


wise interleaving of the coded bits and by using a soft-input soft-output dempper as
the input to the decoder, improves the code diversity and as a result the performance
of the system. In similar fashion to different iterative schemes, iterative decoding can
be applied to BICM, the so called bit-interleaved coded modulation with iterative de-
coding (BICM-ID) [2]. The system model can be seen in Fig. 1. The information bits
q are encoded and then interleaved (Π) to produce the interleaved code bits c. Then
the bits are mapped onto the symbols x. After transmission over the channel, in our

A NACHRICHTENTECHNIK Version 1, September 5, 2014 1


Matlab Project 3 Iterative Methods (WS2014/2015)

case the AWGN channel, the symbols y are received. Using the soft-input soft-output
demapper, the LLR values Ld are produced, de-interleaved Π−1 and then decoded to
produce the estimate of the information bits q̂. In the case of iterative decoding, ex-
trinsic information is exchanged between the decoder and the demapper. They are
calculated as follows:

LDEC = Π(Lcc − LDEM ) (1)

LDEM = Π−1 (Ld − LDEC ) (2)

LDEC is used as a-priori information for the demapper and LDEM is used as the LLR
values for decoding.

1.2 Extended Mapping

110 100
011 001

111 000
010 101

Figure 2: Extended mapping for 4QAM.

Extended mapping schemes employ more than one bit label per signal point. This
increases the label length, resulting in an ambiguous mapping scheme. The extra bits
can be used with a stronger channel code to increase the power efficiency, without de-
creasing the bandwidth efficiency of the scheme, or employed with higher rate channel
code to increase the bandwidth efficiency of the scheme. Using iterative demapping and
decoding, the ambiguity of the bits can be resolved through the exchange of extrinsic
information between the demapper and the decoder.

In Fig. 2, an extended mapping scheme for 4QAM is depicted. Each symbol is as-
signed two labels, with the first and third bit having ambiguity. The mapping can
be optimized by the use of extrinsic information transfer (EXIT) charts, or by the
calculation of cost functions based on the mutual information or other metrics. The
optimization of the extended mapping is not covered in this project.

2 Version 1, September 5, 2014 A NACHRICHTENTECHNIK


Iterative Methods (WS2014/2015) Matlab Project 3

2 Simulation
For the simulation of the system, a skeleton of the code is provided with the exercise.
The convolutional encoder and BCJR decoder are provided as mex functions with
the project. The tasks are to complete the missing parts and in the end, assess the
performance of the system.

2.1 Mapper

The bit-to-symbol mapper takes the input stream of bits c and maps it according to
the given mapping to the symbols of the alphabet A.

ä Matlab Task: Implement the following Matlab function: x=mapBits(c,A,MAP):


Maps the input bits c.
ä Hint: An easy way to define the multiple mappings is to assign each pair of mappings
from the matrix MAP to their respective symbols, i.e. MAP(:,1) and MAP(:,2) to A(1).

2.2 Soft-bit demapper

The demapper produces the soft-bit LLR values according to the following equation:
 Q
−|y−ai |2
P 
m
ai ∈A0j exp 2
σn k=1 exp LDEC
k (1 − ckai )
Ld (cj |y) = log  P  Q  , (3)
−|y−ai |2 m
ai ∈A1j exp 2
σn k=1 exp LDEC
k (1 − ckai )

where y is the received symbol, ai is the transmit symbol, m the number of bits per
symbol and LDEC
k the a-priori LLR value of the bit ckai , the coded bit at the position
k of the symbol ai . The sets A0 and A1 denote the set of symbols where the bit cj is
0 or 1 respectively.

ä Matlab Task: Implement the following Matlab function: L d=softDemap(y, sigma n,


A, MAP, L dec): Calculates the LLR values L d.
ä Hint: Use the matrix MAP to differentiate between the symbols of the sets A0 and
A1 . When no a-priori LLR is available, L dec should be set to zero. One possible
solution is to use NaN (Not-a-Number) to set L dec inside the demapper function.

References
[1] G. Caire, G. Taricco, and E. Biglieri, “Bit-interleaved coded modulation,” Infor-
mation Theory, IEEE Transactions on, vol. 44, no. 3, pp. 927–946, May 1998.
[2] X. Li and J. Ritcey, “Bit-interleaved coded modulation with iterative decoding,”
Communications Letters, IEEE, vol. 1, no. 6, pp. 169–171, Nov 1997.

A NACHRICHTENTECHNIK Version 1, September 5, 2014 3

You might also like