CN101888358B - Transmission method for reducing computational complexity of two-way relay nodes based on network coding - Google Patents

Abstract

The invention discloses a transmission method for reducing computational complexity of two-way relay nodes based on network coding. In two-way cooperative relay communication, the self information of first two time slots is convolutionally coded respectively by two information sources for transmitting the information to a relay; the relay is used for respectively detecting estimation values before decoding; the two-way estimation values are processed through networking coding and are jointly decoded to directly obtain a network coding form; a third time slot relay is used for recoding, modulating and transmitting the obtained information; and a receiving end is used for jointly processing the data of a previous time slot and received data according to the internal mechanism of the network coding to obtain required information. By using the invention, an error rate identical with a two-way relay channel of the network coding in a traditional DF (Decode and Forward) mode is obtained; convolutional decoding operation is only carried out once in the processing of the relay and the computational complexity of convolutional decoding is high, thereby, the computational complexity is reduced by approaching half, the energy consumption of the relay nodes is saved, and the power efficiency of a communication system is enhanced.

Description

Reduce the transmission method of the two-way via node computation complexity of coding Network Based

Technical field

The present invention relates to the communication transmission technology of wireless network encoding and decoding, relate in particular to a kind of transmission method that reduces the two-way via node computation complexity of coding Network Based.

Background technology

The thought of network code is that people such as Ahlswede proposes in the notion of information flow Network Based in 2000, and they have proved that from information-theoretical angle strictness network code can utilize the network existing resource to reach the heap(ed) capacity of network service to greatest extent.Owing to can improve the throughput and the robustness of wireless network significantly, network code has become one of research focus that network information theory attracts most attention.

Network code in fact just is meant that in the multi-casting communication network intermediate node carries out spreading out of after certain encoding process to information again.In traditional trunking traffic network, just the information that receives is amplified forwardings (Amplify and Forward) or decoding forwarding (Decode and Forward) as the relaying of intermediate node, the stay of two nights is through certain decoded mode acquired information then.And through network code, intermediate node can be encoded each data flow that it is received and redispatched away, thereby improves the throughput and the robustness of network.

Though network code originates from the multicast transmission of cable network, along with studying deeply, network code more and more receives people's attention with the research of other technologies, especially in wireless communication system, shows up prominently, and manifests huge application space.In two-way trunk channel, utilize the characteristic of wireless channel broadcasting, via node is handled through the packet from two-way link being carried out network code, and then sends with the form of broadcasting; Thereby corresponding two receiving terminals all can pass through the via node broadcast message in the two-way communication, with from one's body one-period the information combined decoding that sends, utilize the inherent mechanism of network code, obtain required useful information.

In two-way relay system, user terminal sends upward signal through relaying to the base station, and the base station is sent downstream signal through relaying to the user.Supposing does not have direct channel between base station and the user.In the traditional communication, 4 time slots of above process need are accomplished.Be that the base station sends to relaying, relaying is forwarded to the user, and the user send information to relaying, and relaying is forwarded to the base station.

Adopt linear network encoding only to need 3 time slots can accomplish the transmitted in both directions of information; Be that the base station send information to relaying; The user send information to relaying, and suppose relay can correctly be adjudicated these two information, and the information after relaying will be adjudicated is carried out XOR (XOR) back and broadcasted away.After user side is received this broadcast message, it and the information of oneself are carried out XOR can obtain the downlink information that send the base station.Same information with oneself was carried out the uplink information that XOR obtains user's transmission after broadcast message was received in the base station.

Though relaying adopts network code to reduce by a time slot, need carry out twice detection and twice decode operation, computation complexity is still higher.

Fig. 1 has listed the transmission plan of traditional two-way relay transmission scheme and the present invention's proposition.

A, C are information source among Fig. 1, and B is a relaying, and the traditional trunk channel of (1.1) expression needs four time slots.(1.2) be based on the mode that (DF) transmitted in decoding, need three time slots, wherein b _aAnd b _cRepresent that respectively A sends to the bit information of C and the bit information that C sends to A.Relaying B obtains decoding carrying out after the bit information sequence of A and C the network code that mould 2 adds (XOR) and handles, and sends through chnnel coding, modulation, broadcasting again.(1.3) be a kind of, only need two time slots based on amplifying the mode of transmitting (AF).First time slot is that information source A, C send information simultaneously; The second time slot relaying B amplifies forwarding with the superposed signal that receives; Do not carry out other processing; Though this form is directly perceived and the holding time expense is little, because noise has been influenced performance by the secondary amplification in the repeating process, has restricted application prospect.

Summary of the invention

The object of the present invention is to provide a kind of transmission method that reduces the two-way via node computation complexity of coding Network Based, this method has reduced the computation complexity and the time delay of relaying when having kept identical bit error rate performance, improved system throughput.

A kind of transmission method that reduces the two-way via node computation complexity of coding Network Based provided by the invention, this method comprises the steps:

(1) in two-way relay transmission system, establishing current time slots is 3*t+1, t=0, and 1, Obtain baseband signal X after the original information sequence process convolution coding of the first information source A and the BPSK modulation _aThe first information source A is with baseband signal X _aSend through sharp keen fading channel, via node B carries out Maximum Likelihood Detection to the fading signal that receives and carries out hard decision drawing the estimated value V to the coded message of information source A _a, V _a∈ 0,1};

(2), obtain baseband signal X after the original information sequence of second information source C process convolution coding and the BPSK modulation at the 3*t+2 time slot _c, the second information source C is with baseband signal X _cSend through sharp keen fading channel, the decline information that while via node B will receive carries out Maximum Likelihood Detection and carries out hard decision obtaining the estimated value V to the coded message of the second information source C _c, V _c∈ 0,1};

(3), the detected information estimated value of top two time slots is carried out network code, i.e. V=V by via node B at the 3*t+3 time slot _aXOR V _c, V ∈ 0,1}; Then V is carried out convolutional code decoder and obtain overlapped information b=b the estimated value of two information source original information sequence _aXORb _c, via node B sends the modulation of overlapped information recompile again through broadcasting;

(4) first, second information source A, the C demodulation of decoding respectively obtains the overlapped information b=b of the estimated value of two information source original information sequence _aXORb _c, first, second information source A, C are respectively according to self existing information, through b then _c=bXORb _aAnd b _a=bXORb _cObtain required separately information.

The inventive method combines network code and channel-decoding and is applied to two-way trunk channel, and through combined decoding, this method has reduced the computation complexity and the time delay of relaying, has improved system throughput.The present invention is different from three kinds of transmission meanss of (1.1) among Fig. 1, (1.2) and (1.3); Be the transmission plan that adopts following three time slots of DF mode equally; What be different from figure (1.2) is that via node only detects the symbolic information estimated value before obtaining decoding, and not decoding obtains information sequence, the detection information that obtains is carried out the handled of network code; And then combined decoding obtains the network code form that the mould 2 of two-way information sequence adds, i.e. b _aXORb _cThereby, reduced the computation complexity and the processing delay of via node, improve the power efficiency of system.

The present invention carries out xor operation through direct decoding with the information sequence that draws, but will carry out network code through the symbol that detection obtains before the channel-decoding, unites the information that solves XOR then, and relaying has just reduced the process of a channel decoding like this.Because the computation complexity of convolutional code decoder process is very high, relaying reduces by a convolution code decoding and just can reduce near half the calculating power consumption, thereby has improved power utilization, has reduced processing delay.Particularly, the present invention has following technical characterstic:

(1) is different from above-mentioned DF mode lower network coding part and is that the processing mode of the relaying shown in Fig. 1 .2 is respectively the information that receives to be obtained information sequence b through the decoding demodulation earlier _aAnd b _c, then they are carried out XOR and handle, coded modulation broadcasting sends again; And the present invention does not need to solve fully the two-way information sequence, just it is detected the signal before obtaining decoding, and then detection information is carried out network code and merges, and combined decoding obtains b again _aXORb _c

(2) the convolution code decoding method of the present invention's employing, because convolution code is a linear code, combined decoding again can superpose before decoding.Therefore, so long as, all can carry out aforesaid operations based on the encoding and decoding of linear code.So just can save energy consumption, improve power efficiency.

(3) information detects and adopts hard decision (Hard Detection).At first adopt the signal before Maximum Likelihood Detection (ML) method detects, demodulation obtains decoding.Adopt Viterbi hard decoder mode to solve raw information then.

That (4) trunk channel sends is b _aXORb _cThrough the information after the coded modulation, the message part that source end A and C can cancellations self obtains information needed.

Description of drawings

Fig. 1 is the two-way relay transmission scheme of classics and the trunking plan sketch map that the present invention mentions;

Fig. 2 is that information source A, C and relaying B are to the process sketch map of information processing in the two-way trunk channel of classics;

Fig. 3 is the process sketch map of information source A among the present invention, C and relaying B process information;

Fig. 4 is the flow chart of the information processing of two-way relay system of the present invention;

Fig. 5 is the error rate of the present invention and the legacy network encoding operation error rate;

Fig. 6 is the present invention and legacy network encoding process time delay.

Embodiment

The present invention uses the model shown in Fig. 1 .4 to explain.If the information bit data bag b=that information source will be sent (b [1], b [2] ..., b [K]), b [i] ∈ 0,1} represents i information bit, i=1 ..., K, K represent the size of packet.Information bit through coding back (such as convolution code) obtain coded information sequences V=(v [1], v [2] ... V [N]), v [i] ∈ 0, i symbol behind the 1} representative coding, the number of symbol behind the N presentation code.This moment, channel coding rate was K/N, will be coded information sequences V after ovennodulation, obtain baseband signal X, establish and adopt BPSK (two-phase PSK) modulation, then X=(x [1], x [2] ..., x [N]), x [i] ∈ 1,1}.

At the 3*t+1 time slot, t=0,1 ..., the baseband signal X of information source A _aSend to relaying B, then the information Y after the process Rayleigh channel decline that receives of relaying B _aFor

$Y_a={\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="HDA0000023311820000011.png,HDA0000023311820000041.png"\; state="\{\{state\}\}">h</figure-callout>}}_1\&CenterDot;\sqrt{E}\&CenterDot;V_a+\mathrm{Na}---\left(1\right)$

H wherein ₁Be the channel gain of the Rayleigh channel from A to B, E is the transmitting power of information source A, and Na satisfies the circulation symmetry white complex gaussian noise that N (0,1) distributes.While relaying B is Y to received signal _aDetect, adopt Maximum Likelihood Detection mode (ML).If X _a[i] is that 0 or 1 probability is 0.5, and (wherein i representes i symbol, and all i all represent i symbol later on), then maximum a posteriori probability detects and equals Maximum Likelihood Detection.Utilize log-likelihood ratio (LLR) to detect, promptly

$y_a\left[i\right]=\ln \left(\frac{P\left(y_a\right[i\left]\right|x_a\left[i\right]=1)}{P\left(y_a\right[i\left]\right|x_a\left[i\right]=0)}\right)=\ln \left(\frac{P\left(x_a\right[i]=1|y_a\left[i\right])}{P\left(x_a\right[i]=0|y_a\left[i\right])}\right)---\left(2\right)$

Y wherein _aThe individual log-likelihood ratio of coded identification of i of [i] expression information source A.Then according to the BPSK modulation system, to y _a[i] carries out hard decision (H_D, Hard Detection), draws i the symbol v that has encoded of detected information source A _a[i],

$v_a\left[i\right]=\left\{\begin{array}{cc}1,& y_a\left[i\right]>0\\ 0,& y_a\left[i\right]<0\end{array}\right.---\left(3\right)$

At the 3*t+2 time slot, information source C and information source A adopt same coding and modulation system, and be the same with top processing mode, promptly at first utilizes the log-likelihood ratio detection to draw the individual log-likelihood ratio of coded identification of i of information source C:

$y_c\left[i\right]=\ln \left(\frac{P\left(y_c\right[i\left]\right|x_c\left[i\right]=1)}{P\left(y_c\right[i\left]\right|x_c\left[i\right]=0)}\right)=\ln \left(\frac{P\left(x_c\right[i]=1|y_c\left[i\right])}{P\left(x_c\right[i]=0|y_c\left[i\right])}\right)---\left(4\right)$

Carry out hard decision then, draw i the symbol v that has encoded of detected information source C _c[i],

$v_c\left[i\right]=\left\{\begin{array}{cc}1,& y_c\left[i\right]>0\\ 0,& y_c\left[i\right]<0\end{array}\right.---\left(5\right)$

Need to prove the v that solves here _a[i] and v _c[i] all is i symbols after the chnnel coding, do not draw raw information through channel-decoding.In the existing scheme, establishing the channel-decoding function is F, to v _a[i] and v _c[i] carries out channel-decoding respectively and draws original information bits, i.e. b _c[i]=F (V _c[i]) and b _c[i]=F (V _c[i]), then the information bit that obtains is carried out xor operation, obtain XOR form b [the i]=b of two row bit informations _a[i] XOR b _c[i].

Different with existing scheme is that the present invention is here directly to the detected v of coded identification about information source A and C _a[i] and v _c[i] carries out xor operation, gets two and is listed as the XOR form of coded identification sequence

v[i]＝v _a[i]XOR?v _c[i](6)

Be listed as XOR form v [i] integral body of coded identification sequence to two then and carry out channel-decoding, promptly b [i]=F (v [i]) can directly obtain the two row original information bits sequence b of information source A and C _a[i] and b _cThe XOR form b [i] of [i].

Then at the 3*t+3 time slot, via node B utilizes broadcasting to send after the XOR form b [i] of two row bit informations is carried out coded modulation.The judgement of decoding draws the XOR form b [i] of two row bit informations and information source A, C are respectively to the information that receives, then foundation respectively

b _c[i]＝b[i]XOR?b _a[i]

b _a[i]＝b[i]XOR?b _c[i](7)

Obtain information needed separately.

Carried out emulation experiment to above improvement project.The convolution code of (4,1,3) is adopted in the emulation encoding and decoding, and information sequence length is K=1000, and then coding back output sequence length is N=4000.Channel is for satisfying the sharp keen falt fading channel that N (0,1) distributes, and noise is for satisfying Gauss's additive white noise that N (0,1) distributes.Scheme data message after general network code scheme and the improvement is just the same, according to above-mentioned operating procedure, and experimental result such as Fig. 5, shown in Figure 6.

Performance evaluation

Can find out by simulation result; Improved network code scheme (improving NC) has bit error rate much at one with original scheme (General N C); But the average delay of the relay process of new departure is about 4.1; And the relaying average treatment time delay of former scheme is about 2.6, promptly is reduced to originally 63.4%, and equivalence has improved 57.7% for throughput.

Because convolution code is a linear code, combined decoding again can superpose before decoding.Therefore, so long as, all can carry out aforesaid operations based on the encoding and decoding of linear code.So just can save energy consumption, improve power efficiency.

Relaying is not through direct decoding the information sequence that draws to be carried out xor operation, but will carry out network code through the symbol that detection obtains before the channel-decoding, unites the information that solves XOR, and relaying has just reduced the process of a channel decoding like this.Because the computation complexity of convolutional code decoder process is very high, relaying reduces by a convolution code decoding and just can reduce near half the calculating power consumption, thereby has improved power utilization, has reduced processing delay.

What decoding was adopted is the Viterbi hard decoder.For (volume is v) raised the price for n, k, and n representes the output symbol number, and k representes the input bit number, and v representes number of registers.Each time quantum of Veterbi decoding algorithm must carry out 2 ^vIndividual ACS (adding-compare-select) operation, operation of each state, and each ACS relates to 2 ^K+vIndividual addition and 2 ^k-1 binary system relatively.Therefore, the computation complexity of Viterbi is proportional to branch's complexity (branch complexity) 2 of decoding grid ^K+vWhat this paper emulation was adopted is the convolution code of (4,1,3).And the method that the present invention proposes has increased an operation that mould 2 adds, and computation complexity is directly proportional with n.Therefore, when the information sequence endless, the power loss that under the identical situation of the error rate, reduces is near half the in theory.

Claims (1)

1. a transmission method that reduces the two-way via node computation complexity of coding Network Based is characterized in that this method comprises the steps:

(1) in two-way relay transmission system, establishing current time slots is 3*t+1, t=0, and 1, Obtain baseband signal X after the original information sequence process convolution coding of the first information source A and the BPSK modulation _aThe first information source A is with baseband signal X _aSend through rayleigh fading channel, via node B carries out Maximum Likelihood Detection to the fading signal that receives and carries out hard decision drawing the estimated value V to the coded message of information source A _a, V _a∈ 0,1};

(2), obtain baseband signal X after the original information sequence of second information source C process convolution coding and the BPSK modulation at the 3*t+2 time slot _c, the second information source C is with baseband signal X _cSend through Rayleigh (Rayleigh) fading channel, the decline information that while via node B will receive carries out Maximum Likelihood Detection and carries out hard decision obtaining the estimated value V to the coded message of the second information source C _c, V _c∈ 0,1};

(3), the detected information estimated value of top two time slots is carried out network code, i.e. V=V by via node B at the 3*t+3 time slot _aXOR V _c, V ∈ 0,1}; Then V is carried out convolutional code decoder and obtain overlapped information b=b the estimated value of two information source original information sequence _aXORb _c, via node B sends the modulation of overlapped information recompile again through broadcasting;

(4) first, second information source A, the C demodulation of decoding respectively obtains the overlapped information b=b of the estimated value of two information source original information sequence _aXORb _c, first, second information source A, C are respectively according to self existing information, through b then _c=bXORb _aAnd b _a=bXORb _cObtain required separately information.

Images