CN101695016B - Multi-user random access system based on rateless codes and coding and decoding method thereof - Google Patents

Abstract

The invention discloses a multi-user random access system based on rateless codes and a coding and decoding method thereof, which can adaptively adjust the respective transmission code rate by using the adaptive link adaptation characteristic of the rateless codes and the random characteristic when the rateless codes are coded under the conditions that each user transmitter needs not to know channel state information and no feedback retransmission mechanism is needed in the transmission process, and can effectively use the conflict overlapped information of multiple users to simultaneously transmit the information of the multiple users effectively and reliably and detect and separate the information of the multiple users correctly.

Description

Multi-user random access system and coding and decoding method thereof based on no-rate codes

Technical field

The present invention relates to wireless communication field, relate in particular to a kind of multi-user random access system based on no-rate codes and coding and decoding method thereof.

Background technology

In modern mobile radio system, must adopt good multi-user's transmission means to guarantee that a plurality of users transmit information simultaneously effectively reliably.Though traditional ALOHA technology can realize the transmission under multi-user's single channel condition, can not utilize a plurality of user's conflicts or overlapping information, efficient is not high.And traditional CDMA (Code-Division Multiple-Access) transmission means is the information of distinguishing different user by different frequency expansion sequences, to finish the access transmission of multi-user information.IDMA (Interleave-Division Multiple-Access) transmission means that Li Ping proposes (is seen " Interleave DivisionMultiple-Access ", IEEE Trans.Wireless Commun., vol.5, no.4, pp.938-947, Apr.2006.), then be that different user is by using different interleavers, effectively utilize the information of user's aliasing, realize that the signal of different user separates, to finish the access transmission of multi-user information.But no matter be CDMA or IDMA, in the system each user employed be the fixing coded system of traditional code check, before the information of carrying out transmission, need estimate channel parameter according to channel condition information usually, be fixed as the channel error correction coding of R according to code check of channel parameter design.When estimated channel parameter during greater than the channel parameter of reality, though can realize reliable transmission, caused the waste of transmission, because can use the channel error correction coding of higher code check this moment; When estimated channel parameter during less than the channel parameter of reality, then can not realize reliable transmission.In order to make state information that transmitting terminal learns channel with the fixing coded system of design code check, and in transmission course the packet of retransmission of lost, often need to adopt certain feedback mechanism, for example ARQ (AutoRepeat-Request) etc.

Summary of the invention

The objective of the invention is at the deficiencies in the prior art, a kind of multi-user random access system based on no-rate codes and coding and decoding method thereof are provided, need not to know channel condition information at each user's transmitter, and in transmission course, need not under the condition of any feedback retransmission mechanism, can adjust transmission code rate separately adaptively, and can effectively utilize a plurality of users overlapping information of conflicting, realize that the transmission effectively reliably simultaneously of a plurality of user profile and correct detection separate.

The objective of the invention is to be achieved through the following technical solutions: a kind of multi-user random access system based on no-rate codes, formed by transmitting terminal and receiving terminal.Wherein transmitting terminal comprises the transmitter of the individual single channel input of K (K is any positive integer), single channel output, and receiving terminal comprises the receiver of a single channel input, the output of K road.This system runs on the communication two ends.At transmitting terminal, each user has a transmitter, and the structure of each transmitter is all identical.For user k (k=1,2 ..., K), its transmitter comprises encoder-k connected in series, mapper-k and access controller-k:

Encoder-k is used for the message packets of user k is compiled the Rateless Code encoded packets that produces for continually, the data flow of its input termination outer message data bag, and its output is connected to the input of mapper-k;

Mapper-k is used for finishing by the mapping of encoded packets data bit to the encoded packets transmission symbol, and available common linear modulation mapper is realized, the output of its input termination encoder-k, and its output is connected to the input of access controller-k;

Access controller-k mainly finishes the transmission of encoded packets control, determines at random whether present encoding to be wrapped in namely that access channel sends in the current time slots by pre-determined probability, the output of its input termination mapper-k, and its output is connected to channel;

K user's the transmitter access channel that walks abreast is formed the transmitting terminal of system.

At receiving terminal, the receiver of the input of its single channel, the output of K road is divided into parallel K road in inside, and each user correspondence is wherein corresponding one the tunnel, and has identical structure.For user k (k=1,2 ..., K), its corresponding that line structure comprises solution connected in series mapping and soft decoding information estimator-k and judgement decoder-k:

Separate mapping and soft decoding information estimator-k, being used for finishing each subscriber signal according to the aliasing encoded packets of coming self-channel that receives separates, and finish estimation to the soft information of encoded packets bit Log likelihood ratio of user k by separating mapping, it imports the input that comes from channel of termination receiver and all the other respectively adjudicate the soft information output of decoder except judgement decoder-k, and its output is connected to the input of judgement decoder-k;

Judgement decoder-k, be used for finishing the decoding to user k encoded packets, the output of its input termination solution mapping and soft decoding information estimator-k, its court verdict output output decode results, as one tunnel output of receiver, its soft information output be connected to except understanding mapping and soft decoding information estimator-k all the other each separate and shine upon and the input of soft decoding information estimator;

The parallel formation of K line structure receiver, the receiving terminal of composition system.

We abbreviate this multi-user random access system based on no-rate codes as RMA (RatelessMultiple-Access) system.

A kind of coding and decoding method of above-mentioned multi-user random access system based on no-rate codes comprises coding method and interpretation method.At transmitting terminal, the coding of finishing each user according to coding method inserts, and each user's encoded packets access channel is sent; At receiving terminal, according to interpretation method, according to the aliasing encoded packets that receives, the detection of finishing each subscriber-coded bag separates and decoding;

Coding method is as follows: the transmitter of establishing each user all needs to send m message packets, and each message packets is made up of b data bit, comprising a cyclic redundancy check (CRC) code, is used for decoder and judges that whether decoding is successful, uses d _k(i) expression message packets, wherein i be message packets numbering (i=1,2 ..., m), subscript k be the user numbering (k=1,2 ..., K);

Consider wherein k user (k=1,2 ..., K), according to the following steps access of encoding:

1) utilize encoder-k to encode;

With message packets d _k(i) send into encoder-k, by Rateless coded system (as LT Code, RaptorCode etc.), the encoded packets sequence that is produced continually and the Tanner of this sign indicating number figure use v _k(s) encoded packets of presentation code gained, wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K);

Each user independently finishes the cataloged procedure of oneself, use the random number generator that is different from other users during coding, to obtain different with other users, separate Tanner figure, can guarantee the irrelevance of each subscriber signal like this, each user's signal separates during for decoding provides feasibility;

2) utilize mapper-k, carry out by the mapping of encoded packets data bit to the encoded packets transmission symbol;

The encoded packets v that will be obtained by encoder-k _k(s) send into mapper-k, mapping mode can adopt common linear modulation mode (as BPSK, QPSK etc.), finishes by the mapping of encoded packets data bit to the encoded packets transmission symbol: v _k(s) → w _k(s), wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K);

3) utilize access controller-k, carry out the transmission control to encoded packets;

The process mapped coding bag w that will be obtained by mapper-k _k(s) send into access controller-k, any time slot j (j=1,2 ...), user k is with Probability p _kOrder is with w _k(s) access channel sends: use x _k(j) expression user k is at the symbol of time slot j channel transmission, if determine at current time slots j encoded packets w _k(s) access channel is then at encoded packets w _k(s) packet header increases user's beacon information of K bit, k position 1 wherein, and transmission encoded packets w is formed in all the other each positions 0 _k' (s), make x _k(j)=w _k' (s), judge whether w at next time slot _k(s+1) access channel; Otherwise make x _k(j)=0, next time slot continues to judge whether w _k(s) access channel;

Because what adopt is the Rateless coded system, so through the sequence after coding and the mapping w _k={ w _k(1), w _k(2) ..., w _k(s) ... } produce continually, so the sequence that channel sends x _k={ x _k(1), x _k(2) ..., x _k(j) ... } also be endlessly, length is not what fix, x _kEach length corresponding corresponding code check all;

Each user's transmitter produces encoded packets and access channel endlessly according to above rule, informs that until receiving terminal it stops to send.

Interpretation method is as follows:

Receiver receives self-channel, endlessly, the mutual aliasing of each user and added the encoded packets of noise, each user's signal separates and decoding is carried out simultaneously, for user k (k=1,2, ..., K), its judgement decoder-k adopts the random number generator identical with encoder-k, so the Tanner of this sign indicating number of reconstruct schemes accurately, carry out signal separation and decoding according to following steps:

1) receiver at first receives N=m the aliasing encoded packets of coming self-channel;

2) receiver receives Δ N aliasing encoded packets again, makes N=N+ Δ N, and the code check of this moment is $R=\frac{m}{N};$

3) encoded packets that receives is sent into solution mapping and soft decoding information estimator-k, adopt " based on the soft decoding information method of estimation of the multi-user random access system of no-rate codes ", under current code check R, carry out computing, the soft information of output (being 0 during initialization) of utilization judgement decoder of all the other each users except judgement decoder-k is upgraded and is estimated soft information;

4) will upgrade the soft information of estimation that obtains as the input of judgement decoder-k by separating mapping and soft decoding information estimator-k, Tanner figure under the current code check R of reconstruct moves BP (Belief-Propagation) decoding algorithm, finish one and take turns the iterative decoding computing, and upgrade output court verdict information and the soft information of output of judgement decoder-k;

5) utilize the output court verdict information of judgement decoder-k to judge whether decode results satisfies the verification relation that check-node limits among the Tanner figure, perhaps iterations has reached the maximum times that limits, if satisfy one of above-mentioned two conditions, then change step 6) over to; Otherwise change step 3) over to;

6) utilize the cyclic redundancy check (CRC) code judgment data bag in each bag whether all to decipher correctly, if all correct, change step 7) over to; Otherwise change step 2 over to);

7) decoding finishes, and receiver sends a signal and informs that the transmitter of user k stops to send.

The soft decoding information method of estimation of described multi-user random access system based on no-rate codes is as follows:

We establish receiver and receive self-channel, endlessly, each user's aliasing and added that the encoded packets sequence of noise is y={y (1), y (2) ..., y (j) ... }:

$y\left(j\right)=\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{h}_i\·x_i\left(j\right)+n\left(j\right),(j=\mathrm{1,2},...)$

Wherein j is the time slot label, subscript i be the user numbering (i=1,2 ... K), x _i={ x _i(1), x _i(2) ..., x _i(j) ... } be the sequence endlessly that i user's channel sends, h is channel parameter, n=n (1), n (2) ..., n (j) ... } for average is 0, variance is σ ²The white Gauss noise sequence;

At receiver, user's beacon information in the encoded packets packet header that we receive according to arbitrary time slot j, the channel of judging each user of this time slot inserts situation and (a K bit in packet header is made hard decision, k (k=1,2 ..., K) position bit decision result is that k user of 1 expression inserted channel, be that k user of 0 expression do not have access channel), remove user's beacon information then; Use J _kExpression user k (k=1,2 ..., the set of encoded packets access slot K):

J _k＝{j|x _k(j)≠0，j＝1，2，...}＝{J _k(1)，J _k(2)，...，J _k(q _k)，...}

Q wherein _kBe set J _kNumbering (the q of middle element _k=1,2 ...), subscript k be the user numbering (k=1,2 ..., K);

For each user, we will receive the encoded packets sequence of different length at receiver yThe time, namely under different code check R, calculate its output of separating mapping and soft decoding information estimator and estimate soft information, to user k, establish receiver and received N encoded packets y=y (1), y (2) ..., y (N) }, then current code check is R=m/N;

For its access slot set J _kIn each time slot J _k(q _k), the aliasing encoded packets y (J that it receives _k(q _k)), can write:

$y\left(J_k\left(q_k\right)\right)=h_k\·x_k\left(J_k\left(q_k\right)\right)+\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h}_i\·x_i\left(J_i\left(q_i\right)\right)+n\left(J_k\left(q_k\right)\right)$

$=h_k\·x_k\left(J_k\left(q_k\right)\right)+{\mathrm{\ζ}}_k\left(J_k\left(q_k\right)\right),(q_k=\mathrm{1,2},...|J_k\left|\right)$

ζ wherein _k(J _k(q _k)) be at time slot J _k(q _k), other subscriber signals of non-zero and white Gaussian noise sum, we are referred to as to merge noise and be similar to and regard it as Gaussian random variable, and average is designated as E[ζ _k(J _k(q _k))], variance is designated as Var[ζ _k(J _k(q _k))], so can be with received code bag y (J _k(q _k)) also regard as and meet Gaussian characteristics:

y(J _k(q _k))～N(h _k·x _k(J _k(q _k))+E[ζ _k(J _k(q _k))]，Var[ζ _k(J _k(q _k))])，

Wherein $E\left[{\mathrm{\ζ}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\≠k,J_k\left(q_k\right)\∈{J_i,J}_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h}_i\·E\left[x_i\left(J_i\left(q_i\right)\right)\right]$

$\mathrm{Var}\left[{\mathrm{\ζ}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h_i}^2\·\mathrm{Var}\left[x_i\left(J_i\left(q_i\right)\right)\right]+{\mathrm{\σ}}^2$

In separating mapping and soft decoding information estimator-k, at first according to the corresponding average operation rule of selected mapping mode Γ _EWith variance operation rule Γ _Var, by the soft information of bit log likelihood ratio of all the other each subscriber-coded bags, calculate average and the variance of the mapping symbols of all the other each subscriber-coded bags; Then according to the corresponding solution mapping ruler of selected mapping mode F, average and variance by the mapping symbols of the aliasing encoded packets that receives and all the other each subscriber-coded bags of having calculated, according to the Gaussian characteristics that the received code bag meets, estimate the soft information of bit log likelihood ratio of user k encoded packets.

When a time iteration, according to " estimating soft information updating expression formula "

${\mathrm{LLR}}_k^a\left(v\left(q_k\right)\right)=F(y\left(J_k\left(q_k\right)\right),E[{\mathrm{\ζ}}_k\left(J_k\left(q_k\right)\right)],\mathrm{Var}[{\mathrm{\ζ}}_k\left(J_k\left(q_k\right)\right)\left]\right)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h}_i\·E[x_i\left(J_i\left(q_i\right)\right)],$

$\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h_i}^2\·\mathrm{Var}\left[x_i\left(J_i\left(q_i\right)\right)\right]+{\mathrm{\σ}}^2)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h}_i\·{\mathrm{\Γ}}_E\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right),$

$\underset{i\≠k,J_k\left(q_k\right)\∈J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\Σ}}{h}_i^2\·{\mathrm{\Γ}}_{\mathrm{Var}}\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right)+{\mathrm{\σ}}^2)$

(q _k＝1，2，...|J _k|)

Upgrade and estimate soft information, LLR in the formula _k ^a(v (qk)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft dope vector of bit log likelihood ratio of individual encoded packets, L _i ^A-1(v (q _i)) be the q behind the a-1 time iterative decoding of user i _iThe soft dope vector of bit log likelihood ratio of individual encoded packets (during initialization a=1, $L_i^0\left(v\left(q_i\right)\right)=0$ )；

If mapping mode adopts BPSK, with concrete average operation rule Γ _{E, BPSK}, variance operation rule Γ _{Var, BPSK}And solution mapping ruler F _BPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form:

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(c\right)}\right)={2h}_k\&CenterDot;\frac{y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}4{{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

(q _k＝1，2，...|J _k|c＝1，2，...，b)

Y (J wherein _k(q _k)) _(c)Be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives, LLR _k ^a(v (q _k) _(c)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft information of log likelihood ratio of c bit of individual encoded packets, L _i ^A-1(v (q _i) _(c)) be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of c bit of individual encoded packets.

If mapping mode adopts QPSK, then with concrete average operation rule Γ _{E, QPSK}, variance operation rule Γ _{Var, QPSK}And solution mapping ruler F _QPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form:

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{(2c-1)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Re}\left[y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}\right]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}4{{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(2c\right)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Im}\left[y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}\right]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}4{{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

(q _k＝1，2，...|J _k|c＝1，2，...，b/2)

Wherein, y (J _k(q _k)) _(c)Be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives, LLR _k ^a(v (q _k) _(2c-1)), LLR _k ^a(v (q _k) _(2c)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit, L _i ^A-1(v (q _i) _(2c-1)), L _i ^A-1(v (q _i) _(2c)) be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit, Re[.] expression gets real part computing, Im[.] expression gets imaginary-part operation;

To estimate soft dope vector LLR then _k ^a(v (q _k)) as the soft dope vector L of input that adjudicates decoder-k _k ^{In, a}(q _k), even $L_k^{\mathrm{in},a}\left(q_k\right)={\mathrm{LLR}}_k^a\left(v\left(q_k\right)\right),$ Send into the iterative decoding that judgement decoder-k carries out a time; All the other each users are also by this same operation, and each user's decoding is carried out simultaneously; After finishing a time iteration, for user k, obtain the soft dope vector L of output of its judgement decoder-k _k ^{Out, a}(q _k), order $L_k^a\left(v\left(q_k\right)\right)=L_k^{\mathrm{out},a}\left(q_k\right),$ All the other users are also by this same operation, then will be worth substitution accordingly and " estimate soft information updating expression formula ", beginning is the iteration of (the a+1 time) newly once, by that analogy, satisfy verification relation or the iterations that check-node limits among the Tanner figure until the decode results of adjudicating decoder-k and reach the maximum times that limits;

The soft information that can finish like this under the code check R=m/N is estimated, because the Rateless attribute of coded system, its encoded packets sequence length N constantly changes, we can be under different received code packet lengths, be under each code check, move each self-corresponding computing, thereby finish the soft information estimation under each code check, and then finish signal separation and decoding.

The invention has the beneficial effects as follows: the present invention takes full advantage of the adaptive link fit characteristics of no-rate codes, with and when coding stochastic behaviour---the no-rate codes encoded packets of different user generates at random according to certain way, mutual incoherent this characteristic, need not to know channel condition information at each user's transmitter, and in transmission course, need not under the condition of any feedback retransmission mechanism, can adaptive adjustment transmission code rate separately, and can effectively utilize a plurality of users overlapping information of conflicting, realize that the transmission effectively reliably simultaneously of a plurality of user profile and correct detection separate.

Description of drawings

Fig. 1 is the concrete structure block diagram of RMA system;

When Fig. 2 is the employing LT Code of RMA system coding, the Tanner figure of its LT Code

When Fig. 3 was the employing Raptor Code of RMA system coding, the Tanner of the LDPC Code of its outer sign indicating number part schemed;

When Fig. 4 is the employing Raptor Code of RMA system coding, the Tanner figure of the LT Code of its ISN part;

When Fig. 5 is the employing Raptor Code of RMA system coding, the whole Tanner figure of its Raptor Code;

Fig. 6 is the RMA system (each inserts with probability 1, during code check R=1/16 with adopting the BPSK modulation system per family) that adopts Raptor Code and (each the user R of IDMA system that adopts Zigzag Code _Zigzag=1/2,8 times of spread spectrums, code check R=1/16) the bit error rate comparison diagram under each signal to noise ratio, number of users K=2;

Fig. 7 is the RMA system (each inserts with probability 1, during code check R=1/16 with adopting the BPSK modulation system per family) that adopts Raptor Code and (each the user R of IDMA system that adopts Zigzag Code _Zigzag=1/2,8 times of spread spectrums, code check R=1/16) the bit error rate comparison diagram under each signal to noise ratio, number of users K=4.

Embodiment

Followingly with reference to accompanying drawing the present invention is further described, it is more obvious that purpose of the present invention and effect will become.

No-rate codes (Rateless Code) is a kind of novel channel coding method with adaptive link fit characteristics.It is that with the fixing coded system maximum of traditional code check different it does not set cbr (constant bit rate) at transmitting terminal, and the number of its encoded packets is hard-core.Transmitting terminal can produce encoded packets endlessly at random and sends according to certain way.Receiving terminal then can be after receiving these encoded packets trial and error decoding.From the statistical significance, each encoded packets is comprising the identical amount of information about message packets, receiving terminal also is indifferent to the reception condition of concrete some encoded packets, but is concerned about the total quantity of the encoded packets that receives, and therefore need not the mechanism of feeding back and retransmitting in transmission course.As long as receiving terminal receives abundant encoded packets, just can decoding success; If decoding failure, receiving terminal some encoded packets that receive again continues trial and error decoding then.Receiving terminal will repeat this process up to successfully decoded always.In case successfully decoded, receiving terminal only need send a very simple signal and inform that transmitting terminal stops transmission and gets final product, and has so just finished whole transmission course.At this moment, the code check of actual transmissions depends on the encoded packets number of actual transmission, and the actual encoded packets number that sends depends on channel conditions at that time.This shows, adopt the no-rate codes coded system, can not know any channel condition information at transmitting terminal, and need not to adjust transmission code rate adaptively under the condition of any feedback retransmission mechanism in transmission course, guarantee information is transmission effectively reliably.

With reference to Fig. 1, a RMA system that comprises the individual user of K (K is any positive integer) is made of transmitting terminal and receiving terminal.Wherein transmitting terminal comprises the transmitter of K single channel input, single channel output, and receiving terminal comprises the receiver of 1 single channel input, the output of K road.This system runs on the communication two ends.At transmitting terminal, each user has a transmitter, and the structure of each transmitter is all identical.For user k (k=1,2 ..., K), its transmitter comprises encoder-k connected in series, mapper-k and access controller-k:

Encoder-k is used for the message packets of user k is compiled the Rateless Code encoded packets that produces for continually, the data flow of its input termination outer message data bag, and its output is connected to the input of mapper-k;

Mapper-k is used for finishing by the mapping of encoded packets data bit to the encoded packets transmission symbol, and available common linear modulation mapper is realized, the output of its input termination encoder-k, and its output is connected to the input of access controller-k;

Access controller-k mainly finishes the transmission of encoded packets control, determines at random whether present encoding to be wrapped in namely that access channel sends in the current time slots by pre-determined probability, the output of its input termination mapper-k, and its output is connected to channel;

K user's the transmitter access channel that walks abreast is formed the transmitting terminal of system.

At receiving terminal, the receiver of the input of its single channel, the output of K road is divided into parallel K road in inside, and each user correspondence is wherein corresponding one the tunnel, and has identical structure.For user k (k=1,2 ..., K), its corresponding that line structure comprises solution connected in series mapping and soft decoding information estimator-k and judgement decoder-k:

Separate mapping and soft decoding information estimator-k, being used for finishing each subscriber signal according to the aliasing encoded packets of coming self-channel that receives separates, and finish estimation to the soft information of encoded packets bit Log likelihood ratio of user k by separating mapping, it imports the input that comes from channel of termination receiver and all the other respectively adjudicate the soft information output of decoder except judgement decoder-k, and its output is connected to the input of judgement decoder-k;

Judgement decoder-k, be used for finishing the decoding to user k encoded packets, the output of its input termination solution mapping and soft decoding information estimator-k, its court verdict output output decode results, as one tunnel output of receiver, its soft information output be connected to except understanding mapping and soft decoding information estimator-k all the other each separate and shine upon and the input of soft decoding information estimator;

The parallel formation of K line structure receiver, the receiving terminal of composition system.

The coding and decoding method of above-mentioned multi-user random access system based on no-rate codes comprises coding method and interpretation method.At transmitting terminal, the coding of finishing each user according to coding method inserts, and each user's encoded packets access channel is sent; At receiving terminal, according to interpretation method, according to the aliasing encoded packets that receives, the detection of finishing each subscriber-coded bag separates and decoding.

Coding method is as follows: the transmitter of establishing each user all needs to send m=1024 message packets, each message packets is made up of b=128 data bit, comprising a cyclic redundancy check (CRC) code, this cyclic redundancy check (CRC) code adopts CRC16, be used for decoder and judge that whether decoding is successful, uses d _k(i) expression message packets, wherein i be message packets numbering (i=1,2 ..., 1024), subscript k be the user numbering (k=1,2 ..., K);

Consider wherein k user (k=1,2 ..., K), according to the following steps access of encoding:

1) utilize encoder-k to encode;

With message packets d _k(i) send into encoder-k, by Rateless coded system (as LT Code, RaptorCode etc.), the encoded packets sequence that is produced continually and the Tanner of this sign indicating number figure.Use v _k(s) encoded packets of presentation code gained, wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K).

Embodiment 1: when coding adopts LT Code, encode in the following manner:

M=1024 message packets carried out the LT coding, generate the Tanner figure of LT Code, and the encoded packets of the LT Code that is generated continually.Use v _k(s) encoded packets of expression LT Code, wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K), coding method is: produce LT Code encoded packets v _k(s), at first (each parameter of this distribution is as shown in table 1, wherein Ω according to number of degrees distribution Ω (x) _iThe shared ratio of expression number of degrees i), produce number of degrees deg with random number generator for it _k(s); Produce one 1024 dimension two element field vector with random number generator then $\left\{G_{k_{\mathrm{is}}}\right\}(i=\mathrm{1,2},...,1024),$ Element

Value be " 0 " or " 1 ", wherein the number of element " 1 " is deg _k(s).Element

For " 1 " expression corresponding be numbered i message packets d _k(i) selected, the message packets that these are selected is done mould 2 summations by bit, namely obtains the value of the encoded packets of LT Code:

$v_k\left(s\right)=\left[\underset{i=1}{\overset{1024}{\mathrm{\&Sigma;}}}{G}_{k_{\mathrm{is}}}{d}_k\left(i\right)\right]\mathrm{<figure-callout\; id="2"\; label="mod"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">mod</figure-callout>}2,(s=\mathrm{1,2},...)$

Ω 1	Ω 2	Ω 3	Ω 4	Ω 5	Ω 8
						0.007969	0.493570	0.166220	0.072646	0.082558	0.056058
Ω 9	Ω 19	Ω 65	Ω 66	All the other parameters
						0.037229	0.0055590	0.025023	0.003135	0

Table 1

The Tanner figure of LT Code as shown in Figure 2.Two category nodes are arranged among the figure, and circle is represented variable node (variable node), and variable node is divided into two classes again, above a behavior information node (information node), total m=1024, note is made V _{K, 1}, V _{K, 2}..., V _{K, m}, the corresponding message packets d of difference _k(1), d _k(2) ..., d _k(m), below a behavior coding nodes (parity node), note is made P _{K, 1}, P _{K, 2}..., P _{K, s}..., the encoded packets v of the corresponding LT Code of difference _k(1), v _k(2) ..., v _k(s) ...; Box indicating check-node (check node), note is made C respectively _{K, 1}, C _{K, 2}..., C _{K, s}..., the verification of each node that links to each other with each check-node and be 0.Therefore, using random number generator in cataloged procedure is the encoded packets v of each LT Code _k(s) select deg _k(s) individual message packets is done mould 2 summations, just is presented as to be check-node C in Tanner figure _{K, s}Select deg _k(s) individual information node links to each other.And check-node C _{K, s}With coding nodes P _{K, s}Annexation fix C _{K, s}Always with P _{K, s}Link to each other one to one.Because LTCode is a kind of Rateless coding, its encoded packets number is along with cataloged procedure increases endlessly, so new LT Code encoded packets v of every generation _k(s), will increase a new check-node C among the figure _{K, s}With a new coding nodes P _{K, s}, along with encoded packets increases endlessly, Tanner figure can be increasing.

Each user independently finishes the LT Code cataloged procedure of oneself, use the random number generator that is different from other users during coding, to obtain different with other users, separate Tanner figure, can guarantee the irrelevance of each subscriber signal like this, each user's signal separates during for decoding provides feasibility;

Embodiment 2: when coding adopts Raptor Code (it is made up of the LT Code cascade of the high code check LDPC Code of outer sign indicating number part and ISN part), encode in the following manner:

The coding first step, (see " Regular and IrregularProgressive Edge-Growth Tanner Graphs " according to Linear-Time PEG Encoding algorithm, IEEE Transactions On InformationTheory, Vol.51, No.1, January 2005), m=1024 message packets carried out the LDPC coding, generate the Tanner figure of LDPC Code, and then obtain the encoded packets of n=1072 LDPC Code.Use t _k(l) encoded packets of expression LDPC Code, wherein l be encoded packets numbering (l=1,2 ..., 1072), subscript k be the user numbering (k=1,2 ..., K).

The Tanner figure of the LDPC Code of the outer sign indicating number part of Raptor Code has two category nodes as shown in Figure 3 among the figure, circle is represented variable node (variable node), and total n=1072, note is made V _{K, 1}, V _{K, 2}..., V _{K, n}, the encoded packets t of the corresponding LDPC Code of difference _k(1), t _k(2) ..., t _k(n); Box indicating check-node (check node), total n-m=1072-1024=48, note is made C ' respectively _{K, 1}, C ' _{K, 2}..., C ' _{K, n-m}, the connected mode of itself and variable node is generated by Linear-Time PEG Encoding algorithm, the verification of each node that links to each other with each check-node and be 0.Among we the definition of T anner figure, be connected to the number of degrees that add up to this node on the limit of certain node.In the variable node of we fixed L DPC Code, the number of degrees of preceding 5 nodes are 1, and the number of degrees of all the other nodes are 4, for the annexation of each variable node selection with check-node, constitute Tanner figure according to Linear-Time PEG Encoding algorithm.The Tanner that is obtained by this algorithm schemes corresponding check matrix H, has following character: H=[H ^p, H ^d], the H of right-hand component ^dBe 48 * 1024 matrixes, the H of left-hand component ^pBe 48 * 48 upper triangular matrixs:

Then the value of the encoded packets of LDPC Code can be tried to achieve by following formula:

$t_k\left(l\right)=\left\{\begin{array}{cc}[\underset{i=l+1}{\overset{48}{\mathrm{\&Sigma;}}}{h}_{l,i}^p\&CenterDot;t_k\left(i\right)+\underset{i=1}{\overset{1024}{\mathrm{\&Sigma;}}}{h}_{l,i}^d\&CenterDot;d_k\left(i\right)]\mathrm{<figure-callout\; id="2"\; label="mod"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">mod</figure-callout>}2& 1\&le;l\&le;48\\ d_k(l-48)& 48<l\&le;1072\end{array}\right.$

As seen, the LDPC Code that is obtained by this algorithm is systematic code, and in its 1072 encoded packets, back 1024 is exactly message packets.

Second step of coding, the encoded packets of n=1072 LDPC Code is carried out the LT coding, generate the Tanner figure of LT Code, and the encoded packets of the LT Code that is generated continually.Use v _k(s) encoded packets (encoded packets of whole Raptor Code just) of expression LT Code, wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K), coding method is: produce LT Code encoded packets v _k(s), at first according to the number of degrees distribution Ω (x) shown in the table 1, produce number of degrees deg with random number generator for it _k(s); Produce one 1072 dimension two element field vector with random number generator then $\left\{G_{k_{\mathrm{is}}}\right\}(l=\mathrm{1,2}...,1072),$ Element

Value be " 0 " or " 1 ", wherein the number of element " 1 " is deg _k(s).Element

Be the corresponding LDPC Code encoded packets t that is numbered l of " 1 " expression _k(l) selected, the LDPC Code encoded packets that these are selected is done mould 2 summations by bit, namely obtains the value of the encoded packets of LT Code:

$v_k\left(s\right)=\left[\underset{l=1}{\overset{1072}{\mathrm{\&Sigma;}}}{G}_{k_{\mathrm{is}}}{t}_k\left(l\right)\right]\mathrm{<figure-callout\; id="2"\; label="mod"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">mod</figure-callout>}2,(s=\mathrm{1,2},...)$

The Tanner figure of the LT Code of Raptor Code ISN part as shown in Figure 4.Two category nodes are arranged among the figure, and circle is represented variable node (variable node), and variable node is divided into two classes again, above a behavior information node (information node), total n=1072, note is made V _{K, 1}, V _{K, 2}..., V _{K, n}, the packet of corresponding LT Code (is the encoded packets t of LDPC Code this moment respectively _k(1), t _k(2) ..., t _k(n)), below a behavior coding nodes (parity node), note is made P _{K, 1}, P _{K, 2}..., P _{K, s}..., the encoded packets v of the corresponding LT Code of difference _k(1), v _k(2) ..., v _k(s) ...; Box indicating check-node (check node), note is made C respectively _{K, 1}, C _{K, 2}..., C _{K, s}..., the verification of each node that links to each other with each check-node and be 0.Therefore, using random number generator in cataloged procedure is the encoded packets v of each LT Code _k(s) select deg _k(s) encoded packets of individual LDPC Code is done mould 2 summations, just is presented as to be check-node C in Tanner figure _{K, s}Select deg _k(s) individual information node links to each other.And check-node C _{K, s}With coding nodes P _{K, s}Annexation fix C _{K, s}Always with P _{K, s}Link to each other one to one.Because LT Code is a kind of Rateless coding, its encoded packets number is along with cataloged procedure increases endlessly, so new LT Code encoded packets v of every generation _k(s), will increase a new check-node C among the figure _{K, s}With a new coding nodes P _{K, s}, along with encoded packets increases endlessly, Tanner figure can be increasing.The Tanner of whole Raptor Code figure as shown in Figure 5, it is made up of the Tanner figure cascade merging of the Tanner figure of LDPC Code and LT Code.

Each user independently finishes the Raptor Code cataloged procedure of oneself, use the random number generator that is different from other users during coding, to obtain different with other users, separate Tanner figure, can guarantee the irrelevance of each subscriber signal like this, each user's signal separates during for decoding provides feasibility;

2) utilize mapper-k, carry out by the mapping of encoded packets data bit to the encoded packets transmission symbol;

The encoded packets v that will be obtained by encoder-k _k(s) send into mapper-k, mapping mode can adopt common linear modulation mode (as BPSK, QPSK etc.), finishes by the mapping of encoded packets data bit to the encoded packets transmission symbol: v _k(s) → w _k(s), wherein s be encoded packets numbering (s=1,2 ...), subscript k be the user numbering (k=1,2 ..., K);

Embodiment 3: when mapping mode is BPSK, and v _k(s) each bit is mapped as w in _k(s) symbol in, mapping mode is: " 0 →+1,1 →-1 " obtains through mapped coding bag w _k(s).

Embodiment 4: when mapping mode is QPSK, and v _k(s) per two bits are mapped as w in _k(s) symbol in, mapping mode is: " 00 →+1+j, 01 →+1-j, 10 →-1+j, 11 →-1-j ", obtain through mapped coding bag w _k(s).

3) utilize access controller-k, carry out the transmission control to encoded packets;

The process mapped coding bag w that will be obtained by mapper-k _k(s) send into access controller-k, any time slot j (j=1,2 ...), user k is with Probability p _kOrder is with w _k(s) access channel sends: use x _k(j) expression user k is at the symbol of time slot j channel transmission, if determine at current time slots j encoded packets w _k(s) access channel is then at encoded packets w _k(s) packet header increases user's beacon information of K bit, k position 1 wherein, and transmission encoded packets w is formed in all the other each positions 0 _k' (s), make x _k(j)=w _k' (s), judge whether w at next time slot _k(s+1) access channel; Otherwise make x _k(j)=0, next time slot continues to judge whether w _k(s) access channel;

Because what adopt is the Rateless coded system, so through the sequence after Rateless coding and the mapping w _k={ w _k(1), w _k(2) ..., w _k(s) ... } produce continually, so the sequence that channel sends x _k={ x _k(1), x _k(2) ..., x _k(j) ... } also be endlessly, length is not what fix, x _kEach length corresponding corresponding code check all;

Each user's transmitter produces encoded packets and access channel endlessly according to above rule, informs that until receiving terminal it stops to send.

Interpretation method is as follows:

Receiver receives self-channel, endlessly, the mutual aliasing of each user and added the encoded packets of noise, each user's signal separates and decoding is carried out simultaneously, for user k (k=1,2, ..., K), its judgement decoder-k adopts the random number generator identical with encoder-k, so the Tanner of this sign indicating number of reconstruct schemes accurately, carry out signal separation and decoding according to following steps:

1) receiver at first receives N=1024 the aliasing encoded packets of coming self-channel;

2) receiver receives Δ N=256 aliasing encoded packets again, makes N=N+ Δ N, and the code check of this moment is

$R=\frac{m}{N}=\frac{1024}{N};$

3) encoded packets that receives is sent into solution mapping and soft decoding information estimator-k, employing is based on the soft decoding information method of estimation of the multi-user random access system of no-rate codes, under current code check R, carry out computing, the soft information of output (being 0 during initialization) of utilization judgement decoder of all the other each users except judgement decoder-k is upgraded and is estimated soft information;

4) will upgrade the soft information of estimation that obtains as the input of judgement decoder-k by separating mapping and soft decoding information estimator-k, Tanner figure under the current code check R of reconstruct moves BP (Belief-Propagation) decoding algorithm, finish one and take turns the iterative decoding computing, and upgrade output court verdict information and the soft information of output of judgement decoder-k;

5) utilize the output court verdict information of judgement decoder-k to judge whether decode results satisfies the verification relation that check-node limits among the Tanner figure, perhaps iterations has reached the maximum times (being made as 150 times) that limits, if satisfy one of above-mentioned two conditions, then change step 6) over to;

Otherwise change step 3) over to;

6) utilize the cyclic redundancy check (CRC) code judgment data bag in each bag whether all to decipher correctly, if all correct, change step 7) over to; Otherwise change step 2 over to);

7) decoding finishes, and receiver sends a signal and informs that the transmitter of user k stops to send.

By above description to decode procedure as can be known, when the user found that the aliasing encoded packets that receives is not enough to correct decoding, receiver need receive 256 aliasing encoded packets again, carries out the iterative decoding of a new round.When receiver was received N encoded packets, corresponding code rate was:

$R=\frac{m}{N}=\frac{1024}{N}$

In order to represent that conveniently, we use the R reciprocal of code check ^-1Portray the variation of code check:

$R^{-1}=\frac{N}{m}=\frac{N}{1024}$

After N encoded packets of each many reception Δs, code check is changed to Δ R ^-1:

${\mathrm{\&Delta;R}}^{-1}=\frac{\mathrm{\&Delta;N}}{m}=\frac{\mathrm{\&Delta;N}}{1024}$

Δ N is 256 herein, so Δ R ^-1Be 0.25.That is to say that each user is with Δ R ^-1=0.25 is stepping, moves iterative decoding under different code checks, until successfully decoded, send a signal and informs that transmitter stops to send.

Soft decoding information method of estimation based on the multi-user random access system of no-rate codes is as follows:

We establish receiver and receive self-channel, endlessly, each user's aliasing and added that the encoded packets sequence of noise is y={y (1), y (2) ..., y (j) ... }:

$y\left(j\right)=\underset{i=1}{\overset{K}{\mathrm{\&Sigma;}}}{h}_i\&CenterDot;x_i\left(j\right)+n\left(j\right),(j=\mathrm{1,2},...)$

Wherein, j is the time slot label, subscript i be the user numbering (i=1,2 ... K), x _i={ x _i(1), x _i(2) ..., x _i(j) ... } be the sequence endlessly that i user's channel sends, h is channel parameter, n=n (1), n (2) ..., n (j) ... } for average is 0, variance is σ ²The white Gauss noise sequence;

At receiver, user's beacon information in the encoded packets packet header that we receive according to arbitrary time slot j, the channel of judging each user of this time slot inserts situation and (a K bit in packet header is made hard decision, k (k=1,2 ..., K) position bit decision result is that k user of 1 expression inserted channel, be that k user of 0 expression do not have access channel), remove user's beacon information then; Use J _kExpression user k (k=1,2 ..., the set of encoded packets access slot K):

J _k＝{j|x _k(j)≠0，j＝1，2，...}＝{J _k(1)，J _k(2)，...，J _k(q _k)，...}

Q wherein _kBe set J _kNumbering (the q of middle element _k=1,2 ...), subscript k be the user numbering (k=1,2 ..., K);

For each user, we will receive the encoded packets sequence of different length at receiver yThe time, namely under different code check R, calculate its output of separating mapping and soft decoding information estimator and estimate soft information, to user k, establish receiver and received N encoded packets y=y (1), y (2) ..., y (N) }, then current code check is R=m/N=1024/N;

For its access slot set J _kIn each time slot J _k(q _k), the aliasing encoded packets y (J that it receives _k(q _k)), can write:

$y\left(J_k\left(q_k\right)\right)=h_k\&CenterDot;x_k\left(J_k\left(q_k\right)\right)+\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;x_i\left(J_i\left(q_i\right)\right)+n\left(J_k\left(q_k\right)\right)$

$=h_k\&CenterDot;x_k\left(J_k\left(q_k\right)\right)+{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right),(q_k=\mathrm{1,2},...|J_k\left|\right)$

ζ wherein _k(J _k(q _k)) be at time slot J _k(q _k), other subscriber signals of non-zero and white Gaussian noise sum, we are referred to as to merge noise and be similar to and regard it as Gaussian random variable, and average is designated as E[ζ _k(J _k(q _k))], variance is designated as Var[ζ _k(J _k(q _k))], so can be with received code bag y (J _k(q _k)) also regard as and meet Gaussian characteristics:

y(J _k(q _k))～N(h _k·x _k(J _k(q _k))+E[ζ _k(J _k(q _k))]，Var[ζ _k(J _k(q _k))])，

Wherein $E\left[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;E\left[x_i\left(J_i\left(q_i\right)\right)\right]$

$\mathrm{Var}\left[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\mathrm{Var}\left[x_i\left(J_i\left(q_i\right)\right)\right]+{\mathrm{\&sigma;}}^2$

In separating mapping and soft decoding information estimator-k, at first according to the corresponding average operation rule of selected mapping mode Γ _EWith variance operation rule Γ _Var, by the soft information of bit log likelihood ratio of all the other each subscriber-coded bags, calculate average and the variance of the mapping symbols of all the other each subscriber-coded bags; Then according to the corresponding solution mapping ruler of selected mapping mode F, average and variance by the mapping symbols of the aliasing encoded packets that receives and all the other each subscriber-coded bags of having calculated, according to the Gaussian characteristics that the received code bag meets, estimate the soft information of bit log likelihood ratio of user k encoded packets.

When a time iteration, according to " estimating soft information updating expression formula "

${\mathrm{LLR}}_k^a\left(v\left(q_k\right)\right)=F(y\left(J_k\left(q_k\right)\right),E[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)],\mathrm{Var}[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\left]\right)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;E[x_i\left(J_i\left(q_i\right)\right)],$

$\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\mathrm{Var}\left[x_i\left(J_i\left(q_i\right)\right)\right]+{\mathrm{\&sigma;}}^2)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;{\mathrm{\&Gamma;}}_E\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right),$

$\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i^{}\&CenterDot;{\mathrm{\&Gamma;}}_{\mathrm{Var}}\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right)+{\mathrm{\&sigma;}}^2)$

(q _k＝1，2，...|J _k|)

Upgrade and estimate soft information, LLR in the formula _k ^a(v (q _k)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft dope vector of bit log likelihood ratio of individual encoded packets, L _i ^A-1(v (q _i)) be the q behind the a-1 time iterative decoding of user i _iThe soft dope vector of bit log likelihood ratio of individual encoded packets (during initialization a=1, $L_i^0\left(v\left(q_i\right)\right)=0$ )；

If mapping mode adopts BPSK, with concrete average operation rule Γ _{E, BPSK}, variance operation rule Γ _{Var, BPSK}And solution mapping ruler F _BPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form:

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(c\right)}\right)={2h}_k\&CenterDot;\frac{y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

(q _k＝1，2，...|J _k|c＝1，2，...，128)

Wherein, y (J _k(q _k)) _(c)Be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives, LLR _k ^a(v (q _k) _(c)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft information of log likelihood ratio of c bit of individual encoded packets, L _i ^A-1(v (q _i) _(c)) be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of c bit of individual encoded packets.

If mapping mode adopts QPSK, then with concrete average operation rule Γ _{E, QPSK}, variance operation rule F _{Var, QPSK}And solution mapping ruler F _QPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form:

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{(2c-1)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Re}\left[y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}\right]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(2c\right)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Im}\left[y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}\right]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4\mathrm{<figure-callout\; id="2"\; label="h\; i"\; filenames="H2009101533943E00021.png,H2009101533943E00022.png"\; state="\{\{state\}\}">h</figure-callout>}}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

(q _k＝1，2，...|J _k|c＝1，2，...，64)

Wherein, y (J _k(q _k)) _(c)Be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives, LLR _k ^a(v (q _k) _(2c-1)), LLR _k ^a(v (q _k) _(2c)) for a time iterative decoding of user k required by the q that separate to shine upon and soft decoding information estimator-k estimates _kThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit, L _i ^A-1(v (q _i) _(2c-1)), L _i ^A-1(v (q _i) _(2c)) be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit, Re[.] expression gets real part computing, Im[.] expression gets imaginary-part operation.

To estimate soft dope vector LLR then _k ^a(v (q _k)) as the soft dope vector L of input that adjudicates decoder-k _k ^{In, a}(q _k), even $L_k^{\mathrm{in},a}\left(q_k\right)={\mathrm{LLR}}_k^a\left(v\left(q_k\right)\right),$ Send into the iterative decoding that judgement decoder-k carries out a time; All the other each users are also by this same operation, and each user's decoding is carried out simultaneously; After finishing a time iteration, for user k, obtain the soft dope vector L of output of its judgement decoder-k _k ^{Out, a}(q _k), order $L_k^a\left(v\left(q_k\right)\right)=L_k^{\mathrm{out},a}\left(q_k\right),$ All the other users are also by this same operation, then will be worth substitution accordingly and " estimate soft information updating expression formula ", beginning is the iteration of (the a+1 time) newly once, by that analogy, satisfy verification relation or the iterations that check-node limits among the Tanner figure until the decode results of adjudicating decoder-k and reach the maximum times that limits;

The soft information that can finish like this under the code check R=m/N=1024/N is estimated, because the Rateless attribute of coded system, its encoded packets sequence length N constantly changes, we can be under different received code packet lengths, be under each code check, move each self-corresponding computing, thereby finish the soft information estimation under each code check, and then finish signal separation and decoding.

Fig. 6 and Fig. 7, be to adopt the RMA system of Raptor Code and adopt Zigzag Code (to see " Zigzagcodes and concatenated zigzag codes ", IEEE Trans.Inform.Theory special issue oncodes on graphs, vol., IT-47, no.2, pp., 800-807, IDMA system Feb.2001.), the performance comparison diagram when system user is counted K=2 and K=4.Can see, adopt the performance of the RMA system of Raptor Code to be better than the IDMA system that adopts Zigzag Code.

Claims (4)

1. the multi-user random access system based on no-rate codes is characterized in that, it mainly is made of transmitting terminal and receiving terminal; Wherein, transmitting terminal is made up of the parallel access channel of the transmitter that K single channel imported, single channel is exported, and receiving terminal is made up of the receiver that a single channel is imported, export on the K road; Each transmitter is by encoder, and mapper and access controller be composition connected in series successively; Receiver is divided into parallel K road in inside, decoder is connected in series forms for the mapping of each route solution and soft decoding information estimator and judgement, and wherein, K is any positive integer;

Encoder-k is used for the message packets of user k is compiled the no-rate codes encoded packets that produces for continually, the data flow of its input termination outer message data bag, and its output is connected to the input of mapper-k;

Mapper-k is used for finishing by the mapping of encoded packets data bit to the encoded packets transmission symbol, realizes with common linear modulation mapper, and the output of its input termination encoder-k, its output is connected to the input of access controller-k;

Access controller-k mainly finishes the transmission of encoded packets control, determines at random whether present encoding to be wrapped in namely that access channel sends in the current time slots by pre-determined probability, the output of its input termination mapper-k, and its output is connected to channel;

Separate mapping and soft decoding information estimator-k, being used for finishing each subscriber signal according to the aliasing encoded packets of coming self-channel that receives separates, and finish estimation to the soft information of encoded packets bit Log likelihood ratio of user k by separating mapping, it imports the input that comes from channel of termination receiver and all the other respectively adjudicate the soft information output of decoder except judgement decoder-k, and its output is connected to the input of judgement decoder-k;

Judgement decoder-k, be used for finishing the decoding to user k encoded packets, the output of its input termination solution mapping and soft decoding information estimator-k, its court verdict output output decode results, as one tunnel output of receiver, its soft information output be connected to except understanding mapping and soft decoding information estimator-k all the other each separate and shine upon and the input of soft decoding information estimator;

k＝1，2，...，K。

2. the coding method of the described connecting system of claim 1, the coding of finishing each user according to coding method at transmitting terminal inserts, each user's encoded packets access channel is sent, it is characterized in that, this coding method is as follows: establishing each transmitter all needs to send m message packets, and each message packets is made up of b data bit, comprising a cyclic redundancy check (CRC) code, be used for the judgement decoder and judge that whether decoding is successful, uses d _k(i) expression message packets, wherein, i is the numbering of message packets, i=1,2 ..., m; Subscript k is user's numbering, k=1, and 2 ..., K; Consider k user wherein, k=1,2 ..., K; According to the following steps access of encoding:

(1) utilize encoder-k to encode: with message packets d _k(i) send into encoder-k, by no rate coding mode, the encoded packets sequence that is produced continually and the Tanner of this sign indicating number figure use v _k(s) encoded packets of presentation code gained, wherein s is the numbering of encoded packets, s=1,2 ...; Subscript k is user's numbering, k=1, and 2 ..., K; Each user independently finishes the cataloged procedure of oneself, use the random number generator that is different from other users during coding, to obtain different with other users, separate Tanner figure, can guarantee the irrelevance of each subscriber signal like this, each user's signal separates during for decoding provides feasibility;

(2) utilize mapper-k, carry out by the mapping of encoded packets data bit to the encoded packets transmission symbol: the encoded packets v that will be obtained by encoder-k _k(s) send into mapper-k, mapping mode adopts common linear modulation mode, finishes by the mapping of encoded packets data bit to the encoded packets transmission symbol: v _k(s) → w _k(s), wherein s is the numbering of encoded packets, s=1, and 2 ...; Subscript k is user's numbering, k=1, and 2 ..., K;

(3) utilize access controller-k, carry out the transmission of encoded packets control: the process mapped coding bag w that will be obtained by mapper-k _k(s) send into access controller-k, at any time slot j, j=1,2 ...; User k is with Probability p _kOrder is with w _k(s) access channel sends: use x _k(j) expression user k is at the symbol of time slot j channel transmission, if determine at current time slots j encoded packets w _k(s) access channel is then at encoded packets w _k(s) packet header increases user's beacon information of K bit, k position 1 wherein, and transmission encoded packets w is formed in all the other each positions 0 _k' (s), make x _k(j)=w _k' (s), judge whether w at next time slot _k(s+1) access channel; Otherwise make x _k(j)=0, next time slot continues to judge whether w _k(s) access channel; Because what adopt is no rate coding mode, so through the sequence after coding and the mapping w _k={ w _k(1), w _k(2) ..., w _k(s) ... } produce continually, so the sequence that channel sends x _k={ x _k(1), x _k(2) ..., x _k(j) ... } also be endlessly, length is not what fix, x _kEach length corresponding corresponding code check all; Each user's transmitter produces encoded packets and access channel endlessly according to above rule, informs that until receiving terminal it stops to send.

3. the interpretation method of the described connecting system of claim 1 according to the aliasing encoded packets that receives, is finished detection separation and the decoding of each subscriber-coded bag at receiving terminal; This interpretation method is as follows: receiver receives self-channel, endlessly, each user mutual aliasing and added the encoded packets of noise, each user's signal separates and decoding is carried out simultaneously; For user k, k=1,2 ..., K, its judgement decoder-k adopts the random number generator identical with encoder-k, so the Tanner of this sign indicating number of reconstruct schemes accurately, carries out signal separation and decoding according to following steps:

(A) receiver at first receives N=m the aliasing encoded packets of coming self-channel;

(B) receiver receives Δ N aliasing encoded packets again, makes N=N+ Δ N, and the code check of this moment is $R=\frac{m}{n};$

(C) encoded packets that receives is sent into solution mapping and soft decoding information estimator-k, employing is based on the soft decoding information method of estimation of the multi-user random access system of no-rate codes, under current code check R, carry out computing, the soft information of output of utilization judgement decoder of all the other each users except judgement decoder-k, be 0 during initialization, upgrade and estimate soft information;

(D) will upgrade the soft information of estimation that obtains as the input of judgement decoder-k by separating mapping and soft decoding information estimator-k, Tanner figure operation BP decoding algorithm under the current code check R of reconstruct, finish one and take turns the iterative decoding computing, and upgrade output court verdict information and the soft information of output of judgement decoder-k;

(E) utilize the output court verdict information of judgement decoder-k to judge whether decode results satisfies the verification relation that check-node limits among the Tanner figure, perhaps iterations has reached the maximum times that limits, if satisfy one of above-mentioned two conditions, then change step F over to); Otherwise change step C over to);

(F) utilize the cyclic redundancy check (CRC) code judgment data bag in each bag whether all to decipher correctly, if all correct, change step G over to); Otherwise change step B over to);

(G) decoding finishes, and receiver sends a signal and informs that the transmitter of user k stops to send.

4. according to the described interpretation method of claim 3, it is characterized in that in the described step (C), the soft decoding information method of estimation of described multi-user random access system based on no-rate codes is specific as follows:

If receiver receives self-channel, endlessly, each user's aliasing and added the volume of noise

Wherein, j is the time slot label, and subscript i is user's numbering, i=1, and 2 ... K; x _i={ x _i(1), x _i(2) ..., x _i(j) ... } be the sequence endlessly that i user's channel sends, h is channel parameter, n=n (1), n (2) ..., n (j) ... } for average is 0, variance is σ ²The white Gauss noise sequence;

At receiver, user's beacon information in the encoded packets packet header that receives according to arbitrary time slot j, the channel of judging each user of this time slot inserts situation: a K bit in packet header is made hard decision, k position bit decision result is that k user of 1 expression inserted channel, be that k user of 0 expression do not have access channel, k=1,2, ..., K; Remove user's beacon information then; Use J _kThe set of the encoded packets access slot of expression user k, k=1,2 ..., K:

J _k＝{j|x _k(j)≠0，j＝1，2，...}＝{J _k(1)，J _k(2)，...，J _k(q _k)，...}；

Wherein, q _kBe set J _kThe numbering of middle element, q _k=1,2 ...; Subscript k is user's numbering, k=1, and 2 ..., K; .

For each user, we will receive the encoded packets sequence of different length at receiver yThe time, namely under different code check R, calculate its output of separating mapping and soft decoding information estimator and estimate soft information, to user k, establish receiver and received N encoded packets y=y (1), y (2) ..., y (N) }, then current code check is R=m/N;

For its access slot set J _kIn each time slot J _k(q _k), the aliasing encoded packets y (J that it receives _k(q _k)), can write:

$y\left(J_k\left(q_k\right)\right)=h_k\&CenterDot;x_k\left(J_k\left(q_k\right)\right)+\underset{i\&NotEqual;k,J_k\left(q_e\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;x_i\left(J_j\left(q_i\right)\right)+n\left(J_k\left(q_k\right)\right)$

q _k＝1，2，...，|J _k|；

Wherein, ζ _k(J _k(q _k)) be at time slot J _k(q _k), other subscriber signals of non-zero and white Gaussian noise sum, we are referred to as to merge noise and be similar to and regard it as Gaussian random variable, and average is designated as E[ζ _k(J _k(q _k))], variance is designated as Var[ζ _k(J _k(q _k))], so can be with received code bag y (J _k(q _k)) also regard as and meet Gaussian characteristics:

y(J _k(q _k))～N(j _k·x _k(J _k(q _k))+E[ζ _k(J _k(q _k))]，Var[ζ _k(J _k(q _k))])，

Wherein, $E\left[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;E(x_i\left(J_i\left(q_i\right)\right)]$

$\mathrm{Var}\left[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\right]=\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h_i}^2\&CenterDot;\mathrm{Var}\left[(x_i\left(J_i\left(q_i\right)\right)\right]+{\mathrm{\&sigma;}}^2;$

In separating mapping and soft decoding information estimator-k, at first according to the corresponding average operation rule of selected mapping mode Γ _EWith variance operation rule Γ _Var, by the soft information of bit log likelihood ratio of all the other each subscriber-coded bags, calculate average and the variance of the mapping symbols of all the other each subscriber-coded bags; Then according to the corresponding solution mapping ruler of selected mapping mode F, average and variance by the mapping symbols of the aliasing encoded packets that receives and all the other each subscriber-coded bags of having calculated, according to the Gaussian characteristics that the received code bag meets, estimate the soft information of bit log likelihood ratio of user k encoded packets;

When a time iteration, according to " estimating soft information updating expression formula "

${\mathrm{LLR}}_k^a\left(v\left(q_k\right)\right)=F(y\left(J_k\left(q_k\right)\right),E[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)],\mathrm{Var}[{\mathrm{\&zeta;}}_k\left(J_k\left(q_k\right)\right)\left]\right)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;E[x_i\left(J_i\left(q_i\right)\right)],$

$\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h_i}^2\&CenterDot;V{\mathrm{ar}\left[x_i\left(J_i\left(q_i\right)\right)\right]+\mathrm{\&sigma;}}^2)$

$=F(y\left(J_k\left(q_k\right)\right),\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;{\mathrm{\&Gamma;}}_E\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right),$

$\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h_i}^2\&CenterDot;{\mathrm{\&Gamma;}}_{\mathrm{Var}}\left(L_i^{a-1}\left(v\left(q_i\right)\right)\right)+{\mathrm{\&sigma;}}^2)$

q _k＝1，2，...，|J _k|；

Upgrade and estimate soft information, in the formula

For a time iterative decoding of user k required by separating the q that mapping and soft decoding information estimator-k estimate _kThe soft dope vector of bit log likelihood ratio of individual encoded packets,

Be the q behind the a-1 time iterative decoding of user i _iThe soft dope vector of bit log likelihood ratio of individual encoded packets; During initialization a=1,

If mapping mode adopts BPSK, with concrete average operation rule Γ _{E, BPSK}, variance operation rule Γ _{Var, BPSK}And solution mapping ruler F _BPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form:

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(c\right)}\right)={2h}_k\&CenterDot;\frac{y{\left(J_k\left(q_k\right)\right)}_{\left(c\right)}-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4h}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

q _k＝1，2，...，|J _k|，c＝1，2，...，b，b＝128；

Y (J wherein _k(q _k)) (c) be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives,

For a time iterative decoding of user k required by separating the q that mapping and soft decoding information estimator-k estimate _kThe soft information of log likelihood ratio of c bit of individual encoded packets,

Be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of c bit of individual encoded packets;

If mapping mode adopts QPSK, then with concrete average operation rule Γ _{E, QPSK}, variance operation rule Γ _{Var, QPSK}And solution mapping ruler F _QPSKSubstitution " is estimated soft information updating expression formula ", can obtain its concrete expression-form;

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{(2c-1)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Re}{[y\left(J_k\left(q_k\right)\right)}_{\left(c\right)}]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4h}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{(2c-1)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

${\mathrm{LLR}}_k^a\left(v{\left(q_k\right)}_{\left(2c\right)}\right)={2h}_k\&CenterDot;\frac{\mathrm{Im}{[y\left(J_k\left(q_k\right)\right)}_{\left(c\right)}]-\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{h}_i\&CenterDot;(1-\frac{2}{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1})}{\underset{i\&NotEqual;k,J_k\left(q_k\right)\&Element;J_i,J_k\left(q_k\right)=J_i\left(q_i\right)}{\mathrm{\&Sigma;}}{{4h}_i}^2\&CenterDot;\frac{e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}}{{(e^{L_i^{a-1}\left(v{\left(q_i\right)}_{\left(2c\right)}\right)}+1)}^2}+{\mathrm{\&sigma;}}^2}$

q _k＝1，2，...，|J _k|，c＝1，2，...，b/2，b＝128；

Wherein, y (J _k(q _k)) _(c)Be illustrated in time slot J _k(q _k) c symbol among the aliasing encoded packets y that receives,

For a time iterative decoding of user k required by separating the q that mapping and soft decoding information estimator-k estimate _kThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit,

Be the q behind the a-1 time iterative decoding of user i _iThe soft information of log likelihood ratio of the odd bits of individual encoded packets and even bit bit, Re[.] expression gets real part computing, Im[.] expression gets imaginary-part operation;

To estimate soft dope vector then

The soft dope vector of input as judgement decoder-k Even

Send into the iterative decoding that judgement decoder-k carries out a time; All the other each users are also by this same operation, and each user's decoding is carried out simultaneously; After finishing a time iteration, for user k, obtain the soft dope vector of output of its judgement decoder-k

Order

All the other users also by this same operation, then will be worth substitution accordingly and " estimate soft information updating expression formula ", the new iteration once of beginning, i.e. the a+1 time iteration; By that analogy, satisfy verification relation or the iterations that check-node limits among the Tanner figure until the decode results of adjudicating decoder-k and reach the maximum times that limits;

The soft information that can finish like this under the code check R=m/N is estimated, because the no speed attribute of coded system, its encoded packets sequence length N constantly changes, we can be under different received code packet lengths, be under each code check, move each self-corresponding computing, thereby finish the soft information estimation under each code check, and then finish signal separation and decoding.

Images