Note: Descriptions are shown in the official language in which they were submitted.
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
A Soft Demodulating Method for 16QAM in Communication System
Technical Field
The present invention relates to a soft demodulation method, more
specifically, to a soft
demodulation method for 16 Quadrature Amplitude Modulation (16QAM) in a
communication system.
Technical Background
Adaptive Modulation and Coding (AMC) is a link adaptation technique widely
used in
mobile communication system. It adaptively chooses link modulation and coding
to adapt the
fading in the link, thereby increasing the system capacity and improving the
communication
quality.
The modulation methods often applied in AMC strategy are Quadrature Phase
Shift
Keying (QPSK) and 16QAM. Compared with QPSK, 16QAM has higher bandwidth
efficiency (twice as QPSK) but lower power efficiency, which means that in
order to obtain
the same Bit Error Rate (BER), Eb/No (the ratio of each bit power to noise
power density)
required by 16QAM is higher than that required by QPSK, in other words, 16QAM
is more
difficult to be demodulated since the constellation points of 16QAM are denser
than those of
QPSK and the demodulation needs to estimate both phase and amplitude.
There are two methods for demodulation at the receiving end, i.e. hard
decision
demodulation and soft decision demodulation. The main idea of the former is to
hard decide
the bit information corresponding to the input of the modulator when
performing
demodulation, which means that what is input to the decoder is the binary bit
information
after hard decision, and the decoder uses a known codeword structure to decide
the codeword
at the input of the encoder. Since some valuable information may be lost by
the demodulator
during each hard decision, the hard decision is not a good solution. However,
by combining
the coding and modulation together, the demodulator will not send some errors
to the decoder.
2
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
The term of "soft decision" usually means that the decoder only temporarily
estimates various
symbols, by which, some information valuable to the decoder will not be lost.
In general, for
the Eb/No of the signal, soft decision gains 2dB more than hard decision, so
most practical
systems adopt the soft decision.
When performing QPSK modulation, one symbol carries the information of two
bits,
which are mapped to I (in-phase) branch and Q (quadrature) branch
respectively. Soft
demodulation can be realized at the receiving end as long as the in-phase part
of the received
symbol after removing the carrier is mapped to I branch and the quadrature
part is mapped to
Q branch, I branch and Q branch correspond to a real value information of a
binary bit
respectively, and the real value information after soft demodulation is serial-
parallel converted
and sent to the decoder to realize soft decision decoding. However, the soft
demodulation is
more complex for 16QAM, since the 16QAM modulation maps four bits to one
symbol,
among which, two bits are mapped to I branch and the other two to Q; when the
soft
demodulation is performed at the receiving end, the in-phase part of the
received symbol after
removing the carrier corresponds to two-bits information and the quadrature
part to the other
two-bits information, and the constellation amplitudes corresponding to the
symbols are
different.
Figure 1 is a basic block diagram of the typical 16QAM modulation and coding/
demodulation and decoding, wherein after Cyclic Redundancy Check (CRC) bits
are added in
the transmitting block, the transmitting block is input to Turbo coding module
for error
correction encoding (step 101), then physical layer hybrid automatic repeat
request HARQ
(step 102), 16QAM base band modulation (step 103) are performed, and sequently
spectrum
spread processing (step 104) is performed, including channelizing and
scrambling, the base
band signal modulates the carrier signal, and the modulated signal then is
transmitted over the
channel (step 105); after the signal is received by the UE, firstly, the
carrier is removed to
obtain the in-phase and quadrature phase signals which are then de-spreaded
(step 106), the
de-spreaded in-phase and quadrature phase symbols are sent to 16QAM soft
decision
3
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
demodulator for soft demodulation (step 107), and the real value soft
information sequence
llqll29z corresponding to the sent binary bit sequenceljgt'z92 is obtained,
then the real value soft
information sequence is sent to Turbo decoder (step 109) for error correction
decoding after
performing physical layer de-HARQ processing (step 108), and finally the
received bit
sequence corresponding to the sent bit sequence is obtained.
At present, in order to solve this problem, a soft decision demodulation
method of
calculating the input to Turbo decoder is often applied. The main idea of this
method is to
calculate the log likelihood ratio (LLR) of each bit corresponding to the in-
phase and
quadrature components of each constellation point, which means the information
after soft
demodulation is the LLR of the corresponding input bit of the modulator. When
this method is
applied, Carrier Signal to Interference (C/I) may need to be estimated for LLR
of some bits
and the error of C/I may affect the performance of the soft demodulation; In
addition, the
calculation process of LLR is relatively complicated and the hardware is
difficult to be
realized.
Summary of the Invention
The technical problem to be solved in the present invention is to provide a
soft
demodulation method for 16QAM in a communication system, so as to provide a
simply
realizable 16QAM soft demodulation method and conveniently achieve adaptive
modulation
and coding strategy.
In order to solve the above technical problem, the present invention provides
the
technical solution of:
estimating the receiving power of the traffic channel according to the pilot
power and the
power deviation between the traffic channel and pilot channel so as to obtain
the average
power Pa,,e of the 16QAM constellation;
removing the carrier from the received intermediate frequency signals to
obtain in-phase
symbol sequence information I and quadrature symbol sequence information Q;
4
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
determining the different decision segments and the corresponding error
probability
decision curves according to the 16QAM constellation mapping relationship
between the
input binary bit sequence tigI1292 and the constellation's position in I, Q
branches, and based
on this, judging said obtained in-phase symbol sequence information and
quadrature symbol
sequence information in different decision segments by using the corresponding
decision
curves in order to obtain the real value soft information sequence i1q1i2q2
In the present invention, said obtained real value sequence 41rz9z can be
further input
into the decoder for error correction decoding to obtain the received bit
sequence
corresponding to the sent bits.
Said soft demodulation method for 16QAM of the present invention fully
integrates the
benefits of soft decision and hard decision, the algorithm is simple, and the
method is easily to
be realized.
Brief Description of the Drawings
Figure 1 is a basic block diagram of the typical 16QAM modulation and coding/
demodulation and decoding;
Figure 2 is a diagram of constellation maps of QPSK and 16QAM;
Figure 3 is a diagram of the mapping principle of I branch and Q branch of
16QAM;
Figure 4 is a sectional diagram of 16QAM clipped soft decision segments;
Figure 5 is a diagram of the algorithm flow of 16QAM soft decision
demodulation.
Preferred Embodiment of the Invention
The basic idea of the present invention is to apply the Clipped Soft Decision
(CSD)
method combining hard decision and soft decision, fully utilize the advantage
of soft decision
against high uncertainty and the advantage of hard decision of preventing over-
estimations so
as to make the algorithm of CSD soft decision simple and easily realized and
have good
performance.
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
In the following, the present invention will be described in detail by taking
the 16QAM
soft decision in High Speed Downlink Packet Access (HSDPA) as an example.
HSDPA is a new technique offered in R5 protocol by 3GPP to meet the needs for
asymmetrical uplink/downlink data service. It solves the conflict between
coverage and
capacity of the system, increases the system capacity greatly and meets the
needs for high
speed services of users. Compared with R99, HSDPA adopts Adaptive Modulation
and
Coding (AMC) and Hybrid Automatic Repeat Request (HARQ) to perform link
adaptation.
The core algorithm of 16QAM CSD soft decision is to apply segment proportional
method similar to QPSK base band mapping to realize soft decision demodulation
according
to the feature of the four bits 'Iql1z92 at the input of 16QAM modulator
corresponding to the
constellation and the different error probability decision curves of the above
four bits, namely
to perform the soft decision respectively by dividing different decision
segments
corresponding to the above riqit292 (corresponding to the thought of hard
decision) so as to
obtain the four real value soft bit information sequence llql1z92
corresponding to the four bits
at the input of the 16QAM modulator.
Table 1 offers the base band modulation mapping of 16QAM in HSDPA. When 16QAM
modulation is applied, four sequential binary symbols hAiz9'2 are series-
parallel connected to
be 1112 on I branch and 9lq2 on Q branch, then the mapping is performed
according to the
mapping principle of Table 1. It should be noted that the average
constellation power of the
constellation mapped according to Table 1 is exactly 1.
Table 1
i 1 q l i2q2 I branch Q branch
0000 0.3162 0.3162
0001 0.3162 0.9487
0010 0.9487 0.3162
0011 0.9487 0.9487
6
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
0100 0.3162 -0.3162
0101 0.3162 -0.9487
0110 0.9487 -0.3162
0111 0.9487 -0.9487
1000 -0.3162 0.3162
1001 -0.3162 0.9487
1010 -0.9487 0.3162
1011 -0.9487 0.9487
1100 -0.3162 -0.3162
1101 -0.3162 -0.9487
1110 -0.9487 -0.3162
1111 -0.9487 -0.9487
Figure 2 is the constellation maps of QPSK and 16QAM. From the constellation
maps, it
can be clearly figured out that constellation points of QPSK have the same
magnitudes and
different phases, while the constellation points of 16QAM may have different
magnitudes and
different phases, and the constellation points of 16QAM are denser than those
of QPSK,
therefore the demodulation, especially the soft decision demodulation of 16QAM
is more
complicated.
Figure 3 concludes the mapping principle of Table 1. When Il or ql is binary
0, a
positive real value signal is bound to be mapped, and when Il or ql is binary
1, a negative
real value signal is bound to be mapped. The mapping of 12 and q2 is more
complicated.
On the basis of Figure 3, Figure 4 further figuratively represents the
principle of clipped
soft demodulation algorithm. Since the mapping principles of 11 and ql are the
same and
the mapping principles of 12 and q2 are the same, the principle of the 16QAM
CSD soft
decision demodulation will be illustrated by taking h, l2 as examples. From
Figure 4, it can
7
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
be seen that when the in-phase symbol information I is positive, the
corresponding lt tends
to be decided as 0, and the greater the I is, the higher the probability of
right decision for 11
is; when I is negative, the corresponding Z, tends to be decided as 1, and the
less the I is, the
higher the probability of right decision for t1 is; similarly, when the
quadrature symbol
information Q is positive, the corresponding q1 tends to be decided as 0, and
the greater the
Q is, the probability of right decision for q1 is; when the Q is negative, the
corresponding q,
tends to be decided as 1, and the less the Q is, the higher the probability of
right decision for
q, is; Therefore, when segment proportional algorithm is applied to perform
the soft decision,
the above tendency can be reflected.
For i2, when the in-phase symbol information 1>0.9487 or 1<-0.9487, the
corresponding
t2 tends to be decided as 1; when -0.3162<1<0.3162, the corresponding t2 tends
to be
decided as 0; When -0.9487<I<-0.3162 or 0.3162<1<0.9487, whether the
corresponding t2
tends to be decided as I or as 0 is decided by the value of I, the more the I
tends to be 0, the
higher the probability of correctness of i2 being decided as 0 is, and the
more the I tends to
be I or -1, the higher the probability of correctness of 12 being decided as I
is; the same
principle is suitable for q2, when the quadrature symbol information Q>0.9487
or Q<-0.9487,
the correspondingqz tends to be decided as 1; when -0.3162<Q<0.3162, the
correspondingq2
tends to be decided as 0; When -0.9487<Q<-0.3162 or 0.3162<Q<0.9487, whether
the
corresponding42 tends to be decided as 1 or as 0 is decided by the value of Q,
the more the Q
tends to be 0, the higher the probability of correctness of q2 being decided
as 0 is, and the
more the Q tends to be I or -1, the higher the probability of correctness of
q2 being decided
as 1 is.
8
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation Therefore the algorithm of the present
invention performs the soft decision
corresponding to the proportional algorithm of the hard decision and exactly
reflects the
above-mentioned tendency. In soft decision, different soft demodulation
equations are applied
to demodulate the soft information i1q1i2q2 corresponding to ~I9l'2g2 for
different segments.
But from analysis, it can be seen that the CSD demodulation equations can be
combined to
obtain the soft decision demodulation algorithm of Figure 5. In the
corresponding soft
decision demodulation equation of Figure 5, 0.7071 corresponds to the in-phase
or quadrature
component of QPSK base band modulation with average constellation power being
1.
In the following, the detailed flow of 16QAM CSD soft decision demodulation
algorithm
of the present invention will be described according to figure 5.
step 501, estimating the receiving power of traffic channel by UE according to
the pilot
power and the power deviation between the traffic channel and the pilot
channel in order to
obtain the average power of each point in 16QAM constellation;
step 502, separating the symbol information of I branch and Q branch;
step 503, directly applying the equations i, =1 * 0 7071 /( PaVe * 0 3 162)
and
q, = Q* 0 7071 /(j.,.* 0 3162) to calculate for the soft decision of t1 and ql
corresponding to the
binary bit sequence tiqt12q2 input into the 16QAM modulator;
step 504, deciding whether I is positive or negative, if I>0, then turning to
step 505; if
1<0, then turning to step 506;
~ *(1/ P,~)-03162 *
step 505, calculating ~2 by the equation t2 -~1-2 0 9487-0 3162 0 7071
1*
Z1+2*(1+03162 o
step 506, calculating 12 by the equationl - 0 9487-0 3162 J 7o7t
step 507, deciding whether Q is positive or negative, if Q>0, then turning to
step 508; if
Q<0, then turning to step 509;
A ( *(Q/ Pme)-03162
q2 -I'1-2 0 9487-0 3162 0 7071
q
step 508, calculating 2 by the equation
9
CA 02625111 2008-04-09
PCT/CN2005/001708 English Translation
~ gZ =(1+2* (Q~~' "')+0 3
q 162)*0 7071
step 509, calculating 2 by the equation l 0 9487-0 3162
step 510, incorporating q, and q2 after soft demodulation to be the real value
~ A A A
sequence 11q112q2 corresponding to bit sequence ilql 12q2 input into the 16QAM
modulator;
step 511, inputting the real value sequence into the decoder for error
correction decoding
to obtain the received bit sequence corresponding to the sent bits.
