CN102820952A - Method for optimizing throughput of wireless network system based on no-rate encoding - Google Patents

Abstract

The invention discloses a method for optimizing the throughput of a wireless network system based on no-rate encoding. In a two-layer transmission system, a no-rate LT code of a network layer is used on an LDPC (low-density parity-check) code of a physical layer. The code generated by the LT encoder is firstly divided into frames with limited lengths, and then all the frames are encoded by the LDPC (low-density parity-check) encoder in sequence. Channel noise and a fading effect can interfere the transmission of the code of the physical layer. On the receiving end, a packet decoded by the physical layer is thrown away if the packet has a fault, namely, only the rightly decoded packet can be transmitted to the network layer. The invention researches the balance relation among various code rates of the network layer and the physical layer on the aspects of network code length and physical layer encoding and provides a solution for optimizing the throughout of the whole system.

Description

Radio Network System throughput optimization method based on no rate coding

Technical field

The present invention relates to field of wireless, the Radio Network System throughput optimization method of particularly encoding based on no rate.

Background technology

A kind of known coded system is the fountain sign indicating number in the no rate coding, and it is a kind of desirable solution that proposes to the application characteristic of large-scale data distribution and reliable broadcast, receives the concern of industrial circle day by day, has obtained the many application of Mi Yue more.The fountain sign indicating number can produce unlimited coded data packet and be used for decoder recovery source data.At the coding transmitting terminal, after receiving that receiving terminal is sent out the affirmation information of rice, coded data packet just stops to generate.For the binary system deletion passage with fixing, known deletion rate, no rate LT coding and Raptor sign indicating number are two groups of most popular fountain sign indicating numbers.As a solution of application layer forward error correction coding, the Raptor sign indicating number has become a kind of standard, supports the file download service of 3GPP multi-medium broadcast/group broadcast service (MBMS) with rice.

In order under wireless channel conditions, to use the notion of fountain sign indicating number, need carry out channel coding rice to anti-fading and noise in physical layer.For example, the two-layer forward error correction coding that all uses of MBMS system: physical layer is used Turbo code (also being a kind of of fountain sign indicating number), and application layer is used the Raptor sign indicating number.Document [M.Vehkapera and M.Medard; " A throughput-delay trade-off in packetized systems with erasures, " in Proc.of IEEE ISIT ' 05, Adelaide; Australia; Sep.2005, pp.1858-1862.] in the probability of erasure of hypothesis physical layer fix, the notion and the throughput information theoretical research framework of multi-level transmission system coding proposed.Document [R.Gowaikar, A.Dana, R.Palanki; B.Hassibi and M.Effros, " On the capacity of wireless erasure relay networks, " in Proc.of IEEE ISIT ' 04; Chicago, USA, Jun.27-Jul.22004; P.400.] when the connection of not considering between physical layer and the network layer, studied the multi-level communication system of the information theory that diminishes packet network.

Because the time-varying characteristics of radio channel need be united the coded system of considering network layer and physical layer.In network layer, represent decoding failure probability with δ rice.Along with the difference that is provided with of δ, accomplish the length L of the coded data packet of decoding operation in required the guaranteeing of receiving terminal _cBe different.The decoding bag that each is attaching header and Cyclical Redundancy Check (CRC) information has formed the input of LDPC (low density parity check code) encoder of a finite length.See different code check R from the physical layer angle _cThe LDPC encoder can produce the code word of different length.Obviously, a balance to be arranged between throughput of system and transmission reliability.R _cSpeed is low more, and the system transmissions reliability is high more but throughput also can be low more.When considering network and physical layer system throughput, R _cWith these two parameters of δ all be very important.

In view of this, look for the solution that improves throughput of system to greatest extent, make the data of transmission successfully received decoding, become a kind of demand of reality.

Summary of the invention

The object of the present invention is to provide a kind of Radio Network System throughput optimization method based on no rate coding; To network layer and physical layer co-design data packet transmission system; Optimize every layer of coding optimal parameter and two-layer between balance, to realize the two-layer optimal throughput performance that the code word size constrained system is all arranged.

Technical scheme of the present invention is following:

A kind of Radio Network System throughput optimization method based on no rate coding comprises the steps:

1) single transmitter sends information to single or a plurality of receiving terminals, and information symbol is divided into Frame, and each Frame comprises K source symbol S={s ₁, s ₂..., s _K;

2) through the LT encoder said Frame is encoded, producing length continuously is L _cCoded data packet;

3) add header P for each coded data packet _hSymbol and CRC P _CRCSymbol forms length L _s=L _c+ P _h+ P _CRCCoded data packet;

4) pass through the LDPC encoder with R _cEncoding rate is L with said length _sThe coded data packet encoder, obtain length N _t=L _s/ R _cLinear block codes word C, said code word C carries out the radio channel transmission through BPSK modulation;

5) for each Frame, transmitter constantly sends coded data packet, receives that up to transmitter rice just stops to send from the affirmation information of all predetermined receiving terminals or the maximum coded data packet quantity that reached permission, sends next Frame then;

6) carry out LDPC decoding at each receiving terminal, any code word that receives in GF (2) territory

does not satisfy condition all will be considered to invalid and abandon whole bag;

7) detect carrying out CRC,, then abandon whole bag if can't pass this detection through the data of LDPC decoding;

8) will send to the LT decoder through the correct decoded data that CRC detects, carry out the LT decoding;

9) after receiving terminal is decoded successfully, send confirmation to transmitter.

Its further technical scheme is: the throughput T of said Radio Network System _sFor

$T_S=\frac{K}{N_t}\·R_{\mathrm{LDPC}}\·R_{\mathrm{LT}},$

R wherein _LDPCBe the correct Solution code rate of LDPC decoder, R _LTIt is the ratio that the information of LT decoder successfully receives.

Its further technical scheme is: in the network layer of Radio Network System, adopt powerful orphan's degree LT coding to distribute, the degree distribution function does

D wherein _MaxBe maximum degree during powerful orphan's degree distributes, ρ _jBe percentage with check-node of j degree,

Said degree distribution function is converted into

S wherein _MaxBe the degree of maximum variable node, s _iBe the ratio of the variable node of degree of having i, be expressed as

$s_i=\frac{e^{-\mathrm{\λ}}{\mathrm{\λ}}^i}{i!},$

The length L of encoded packets _cUnder the probability of 1-δ, satisfy L _c=KZ, wherein Z=∑ _dρ (d)+τ (d), ρ (d) and τ (d) are:

ρ(1)＝1/K

$\mathrm{\ρ}\left(d\right)=\frac{1}{d(d-1)},$ d＝2，3，…，K

$\mathrm{\&tau;}\left(d\right)=\left\{\begin{array}{c}\frac{S}{K}\frac{1}{d},d=\mathrm{1,2},...(K/S)\\ \frac{S}{K}\log (S/\mathrm{\&delta;}),d=K/\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="HDA00001967250500012.png,HDA00001967250500022.png"\; state="\{\{state\}\}">S</figure-callout>}\\ 0,d>K/S\end{array}\right.$

Wherein,

C is a real number between 0 and 1; δ representes probability of decoding failure; Code word size behind network layer LT coding is L _c=K+2log _e(S/ δ) S.

Its further technical scheme is: in the physical layer of Radio Network System, adopt the array code, the check matrix H of array code is by three parameter determining: prime number p and integer k and j, k, j≤p; K and j be in the H matrix in the row and column element be 1 number; H is the matrix of jp * kp, is expressed as:

Wherein I is p * p cell matrix, and α is the single left side of representative or the permutation matrix of right cyclic shift;

Code word size N behind the LDPC coding _tLength K with message part in the code word _LDPCBe respectively N _t=k * p and K _LDPC=(k-j) * p; Encoding rate is R _c=K _LDPC/ N _tThe code word size that generates in physical layer LDPC coding back is:

$N_t=\frac{(K+2{\log }_e(S/\mathrm{\&delta;})S+P_{\mathrm{CRC}}+P_h)}{R_c}.$

Its further technical scheme is: work as R _cIn the time of fixedly, the encoded packets length L _c=K+2Slog _e(S/ δ), wherein S is the monotonically decreasing function of δ, K is the length of source information piece, L _cBe the monotonically decreasing function of δ,

$L_c=K+2\left({\log }_e(K/\mathrm{\&delta;})\sqrt{K}\right){\log }_e\left(\frac{{\log }_e(K/\mathrm{\&delta;})\sqrt{K}}{\mathrm{\&delta;}}\right),$

L _cThe function that is expressed as δ is L _c=f ₁(δ), the throughput T of system _sDepend on K, N _t, R _LTAnd R _LDPC, N wherein _t=L _c/ R _c, R _LTAnd R _LDPCBe the ratio of every layer of successful decoded information, R _LTWith the relation function of δ be R _LT=a ^δ, a ∈ (0,1) wherein;

R _LDPCValue near 1, T then _s(δ)=f ₂(L _c, δ, a ^δ)

$T_s\&ap;\frac{{\mathrm{KR}}_c}{L_c}{\mathrm{\&alpha;}}^{\mathrm{\&delta;}},$

Being expressed as function is T _s(δ)=f ₂(L _c, δ, a ^δ), the maximum throughput of Radio Network System is maxT _s(δ)=f ₂(L _c, δ, α ^δ), δ＞0;

Work as R _cWhen variable, physical layer code check R _cFor

$R_c=\frac{K_{\mathrm{LDPC}}}{N_t}=\frac{(k-j)\&times;p}{k\&times;p}=1-\frac{j}{k},$

Wherein, R _cBy j and k decision, R _LDPCBe expressed as R _LDPC=f ₃(j), R _LDPCIt is the dull increasing function of j; K _LDPCBe expressed as K _LDPC=f ₄(j), K _LDPCIt is the monotonically decreasing function of j; The ratio R of the correct decoding of the information of network layer _LTBe the monotonically decreasing function of j, be expressed as R _LT=f ₅(j); The maximum throughput of Radio Network System is maxT _s(j)=f ₆(K, R _LT, R _LDPC), j>=2.

Its further technical scheme is: when δ=0.001, the throughput of whole Radio Network System is maximum.

Its further technical scheme is: work as SNR=10dB, j=6, corresponding R _c=0.793 o'clock, the throughput of whole Radio Network System was maximum.

Useful technique effect of the present invention is:

The present invention is through layer design of striding of network layer and physical layer; The various parameters of phase-split network layer and physical layer are to the influence of throughput of system; Draft the variety of issue of optimization, and propose solution, realized the throughput of system optimization under network layer and the limited situation of physical layer code word size.

Description of drawings

Fig. 1 is the structure chart of Frame and packet.

Fig. 2 is the system model structure chart.

Fig. 3 is block of information length K=500 o'clock, required L _cThe histogram of individual symbol decoding.

Fig. 4 is R _LTChange curve with δ.

Fig. 5 is when K=500, and the balance between physical layer LDPC coding and the network layer LT coding is with the change curve of the setting of δ.

Fig. 6 is when K=500, and the balance between physical layer LDPC coding and the network layer LT coding is with R _cThe change curve of setting.

Embodiment

Further specify below in conjunction with the accompanying drawing specific embodiments of the invention.

System model

Data packet transmission system model of the present invention is following: a single transmitter will send to single or a plurality of users with information.Information symbol is divided into Frame.Each Frame comprises K source symbol: S={s ₁, s ₂..., s _K.The LT encoder is L with their codings and continuous generation length _cCoded data packet.Each packet is coupled with header (P subsequently _hSymbol) and cyclic redundancy check (CRC) (P _CRCSymbol) forming a length is L _sPacket, L _s=L _c+ P _h+ P _CRCThen, the LDPC encoder is with R _cEncoding rate is encoded it, and obtaining length with this meter is N _tLinear block codes word C, N _t=L _s/ R _cSubsequently, code word C is through using the transmission of BPSK (biphase phase shift keying becomes analog signal conversion one of conversion regime of data value) modulation carrying out radio channel.Fig. 1 is the structure chart of Frame and packet.

For each Frame, transmitter constantly sends packet, receives that up to it rice just stops not send out from the affirmation of all predetermined receiving terminals or the bag that reached the maximum quantity of permission.And then give Frame.At receiving terminal; Each user is earlier with the LDPC codeword decoding, and any code word that receives in GF (2) territory

does not satisfy condition

all will be considered to invalid and be dropped.But, some mistake that is not detected also satisfies this test condition.Wherein will carry out CRC through the data of LDPC decoding smoothly and detect,, abandon whole bag if can't pass this detection.Otherwise correct decoded data just send to the LT decoder, thereby with rice LT decoding.After receiving terminal is decoded successfully, can send a confirmation to transmitter.System diagram is as shown in Figure 2.

Qualification among the present invention is following:

1, physical layer channel is the accurate sorrow channel model that falls, and the sorrow coefficient h that falls remains unchanged in same packet, but the variation of different pieces of information parlor is independently.

2, suppose that all communication channels spatially are independently, do not disturb between the network channel.

3, do not have undetected mistake for guaranteeing, suppose that the wrong detection based on CRC is accurately.

4, the calculating of throughput does not comprise the header and the CRC check symbol of each packet, and it is inappreciable that the redundancy of their band rice is compared with a large amount of load loads.

Be the quantization system performance, calculate the throughput of whole system according to the average rice of propagating symbol in each passage:

$T_S=\frac{K}{N_t}\&CenterDot;R_{\mathrm{LDPC}}\&CenterDot;R_{\mathrm{LT}},---\left(1\right)$

R _LDPCIt is the correct Solution code rate of LDPC decoder; R _LTIt is the ratio that the information of LT decoder successfully receives.

The optimization of throughput of system

The present invention studies the trade-off problem between physical layer and the network layer, adopts and strides layer design, confirms that at every layer different parameter rice improves the throughput of system to greatest extent.

A. the coding of network layer

In network layer, distribute with powerful orphan's degree (RSD) LT sign indicating number, the degree d of the prefabricated detection node among the RSD has determined to produce the source symbol quantity of code-word symbol, and the degree distribution function is:

$d\left(x\right)={\mathrm{\&Sigma;}}_{j=1}^{d_{\max }}{\mathrm{\&rho;}}_j{x}^j.$

D wherein _MaxIt is maximum degree during RSD distributes.ρ _jBe percentage with check-node of j degree,

${\mathrm{\&Sigma;}}_{j=1}^{d_{\max }}{\mathrm{\&rho;}}_j=1,{\mathrm{\&rho;}}_j\&Element;\left(\mathrm{0,1}\right).$

Owing to be from s ₁, s ₂..., s _K-1, s _KEvenly select d symbol rice to form the coded identification of LT coding codeword in (shown in Figure 1), the variable degree distribution function s (x) of node follows Poisson distribution.This degree distribution function can be expressed as s _MaxIt is the degree of maximum variable node.s _iBe the ratio of the variable node of degree of having i, be expressed as

$s_i=\frac{e^{-\mathrm{\&lambda;}}{\mathrm{\&lambda;}}^i}{i!}.$

For guaranteeing correct decoding, the length L of encoded packets _cMust under the probability of 1-δ, satisfy L at least _c=KZ, Z=∑ here _dρ (d)+τ (d), ρ (d) and τ (d) be as follows:

ρ(1)＝1/K

$\mathrm{\&rho;}\left(d\right)=\frac{1}{d(d-1)},d=\mathrm{2,3},...,K---\left(2\right)$

$\mathrm{\&tau;}\left(d\right)=\left\{\begin{array}{c}\frac{S}{K}\frac{1}{d},d=\mathrm{1,2},...(K/S)\\ \frac{S}{K}\log (S/\mathrm{\&delta;}),d=K/\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="HDA00001967250500012.png,HDA00001967250500022.png"\; state="\{\{state\}\}">S</figure-callout>}\\ 0,d>K/S\end{array}\right.---\left(3\right)$

Here

wherein c be a real number between 0 and 1.

Fig. 3 provides several under the given situation that parameter δ is set, and the LT decoder is accomplished the histogram of the required actual data packet quantity of decoding.Code word size behind the LT coding can be expressed as L _c=K+2log _e(S/ δ) S.CRC detects and the additional back formation of header length is L _sCode word.The coded identification length L _sThe input character that is equivalent to a LDPC encoder.Being defined as Linear Mapping to two step encoding schemes forms:

$K\stackrel{\mathrm{LT}}{\&RightArrow;}{L}_c\stackrel{\mathrm{CRC}}{\&RightArrow;}{L}_s\stackrel{\mathrm{LDPC}}{\&RightArrow;}{C}_{N,t}.$

Do not considering owing under the situation of excessive obstruction that causes of internal memory and packet loss, the throughput of two factor affecting systems is arranged: the first, the error of transmission of physical layer, the code word of sending is deleted probably.The second, the packet of reception can't the LT decoding in network layer owing to once lack coded identification.

B. the coding of physical layer

The code check R of the LDPC coding of physical layer _cInfluence for the LT of network layer coding.R _cBelong to R _c∈ [0.5,0.897].For adapting to the different code checks of LDPC sign indicating number, adopt the array code.The array code has very outstanding performance, and its code has Algebraic Structure.When the array code when binary code angle rice is seen, their check matrix has sparse property, thereby can utilize sum-product algorithm (SPA) to its decoding.

The check matrix H of array code is by three parameter determining: prime number p and two integer k and j, k here, j≤p.K and j be in the H matrix in the row and column element be 1 number.H is the matrix of a jp * kp, can be expressed as:

I is p * p cell matrix, and α is the single left side of representative or the permutation matrix of right cyclic shift.According to these parameters, the LDPC code word size N behind the coding _tLength K with message part in the code word _LDPCCan be expressed as respectively: N _t=k * p and K _LDPC=(k-j) * p.Encoding rate is R _c=K _LDPC/ N _tThe final code word size that generates in physical layer is:

$N_t=\frac{(K+2{\log }_e(S/\mathrm{\&delta;})S+P_{\mathrm{CRC}}+P_h)}{R_c},---\left(5\right)$

C. interlayer balance

Find out δ and R from formula (5) _cDetermined N _tThe LT code word size of δ decision network layer, the LDPC encoder is subsequently with its coding.When the size of sending bag is confirmed, R _cInformation symbol length in the decision code word.Different R _cThe performance that can influence LDPC and whole system is set.Therefore need unite these parameter rice of adjustment and reach the optimization effect between two-layer.

1) fixing R _cOptimized throughput: at first, consider the R of all channels _cFix.Partly obtain L from above-mentioned A _c=K+2Slog _e(S/ δ), S are the monotonically decreasing functions of δ.K is the length of source information piece, L _cAlso be the monotonically decreasing function of δ,

$L_c=K+2\left({\log }_e(K/\mathrm{\&delta;})\sqrt{K}\right){\log }_e\left(\frac{{\log }_e(K/\mathrm{\&delta;})\sqrt{K}}{\mathrm{\&delta;}}\right).$

L _cBe expressed as the function of δ, L _c=f ₁(δ).The throughput T of system _sDepend on K, N _t, R _LTAnd R _LDPC, shown in (1).Wherein, N _t=L _c/ R _c, R _LTAnd R _LDPCIt is the ratio of every layer of successful decoded information.According to simulation result, R _LTWith the relation function of δ be: R _LT=a ^δ, wherein a ∈ (0,1) is as shown in Figure 4.

In addition, the performance of LDPC physical layer depends on the length N of code word _tDifferent δ produces different L _cAnd N _tBecause N _tNot significantly difference, R _LDPCValue under sorrow falls channel is more or less the same, and is as shown in table 1.

δ	L c	N t	R LDPC
				0.001	700	1400	0.9779
0.01	638	1276	0.965
				0.1	587	1174	0.962
0.5	557	1114	0.958
				0.9	549	1098	0.953

Table 1 is the R that changes according to variable δ _LDPCValue, simulation result is to obtain 4 decibels fading channel.

Because R _LDPCValue near 1, can be reduced to by wushu (1): T _s(δ)=f ₂(L _c, δ, a ^δ)

$T_s\&ap;\frac{{\mathrm{KR}}_c}{L_c}{\mathrm{\&alpha;}}^{\mathrm{\&delta;}},---\left(6\right)$

Write as the form of function, T _s(δ)=f ₂(L _c, δ, a ^δ).Optimization problem just becomes like this

maxT _s(δ)＝f ₂(L _c，δ，α ^δ)，δ＞0， (7)

This equality shows will obtain maximum T _s, parameter δ is unique optimizable variable.Need to confirm that only δ rice satisfies this optimization.Find optimum value δ=0.001 with binary search algorithm.

2) R _cOptimized throughput under the variable situation: consider R now _cSituation as optimization variable.Physical layer code check R _cCan be expressed as:

$R_c=\frac{K_{\mathrm{LDPC}}}{N_t}=\frac{(k-j)\&times;p}{k\&times;p}=1-\frac{j}{k},---\left(8\right)$

Shown in (8), R _cBy j and k decision.For an array code, if k is certain, adjustment j will obtain one group of R _cIncrease the value of j, R _cJust diminish R _LDPCValue will increase.This is because in communication theory, R _cMore little, just mean that the length that information is encoded is long more, that is to say that information can be big more by the probability of correct decoding at receiving terminal by well protection, just R _LDPCJust big more.R _LDPCHere represent by the ratio of correct decoding.As the function of j, R _LDPCCan be expressed as R _LDPC=f ₃(j), it is the dull increasing function of a j.

For a length is N _tCode word C, R _cMore little, the message length K in the code word C _LDPCJust short more.K _LDPCBe a function, can be expressed as K about j _LDPC=f ₄(j), it is the monotonically decreasing function of j.Table 2 demonstrates the performance performance that is coded in rayleigh fading channel based on the LDPC of array code.

p	j	k	N t	R c	K LDPC	R LDPC
							29	3	29	841	0.897	754	0.1129
29	4	29	841	0.862	725	0.4768
							29	5	29	841	0.828	696	0.717
29	6	29	841	0.793	667	0.8136
							29	7	29	841	0.759	638	0.8526
29	8	29	841	0.724	609	0.8648

The R of table 2 _LDPCValue be to obtain at 9 decibels Rayleigh fading channel.

From what has been discussed above can draw, along with the increase K of j _LDPCWill reduce.

Can find out from formula (8), along with the increase of j, K _LDPCCan reduce.K _LDPCMore little, mean that the code word that the LT encoder goes out rice need play the bag of a plurality of more LDPC, the LDPC bag is many more, and is can success just little the possibility of these LDPC bag reduction at the LT decoder, just R _LTReduce.Therefore, the ratio R of the correct decoding of the information of network layer _LTBe the monotonically decreasing function of j, can be expressed as R _LT=f ₅(j), table 3 has provided simulation result.

	δ	j＝3	j＝4	j＝5	j＝6	j＝7	j＝8
								R LT	0.01	0.999	0.993	0.983	0.933	0.706	0.292
R LT	0.05	0.997	0.991	0.982	0.953	0.859	0.583
								R LT	0.2	0.998	0.996	0.984	0.979	0.932	0.828
R LT	0.5	0.998	0.996	0.984	0.970	0.932	0.853
								R LT	0.9	0.998	0.996	0.984	0.970	0.932	0.853

Table 3, when δ changes, the relative R of j _LTVariation.

Based on above analysis, the optimized throughput problem of Physical layer just becomes

maxT _s(j)＝f ₆(K，R _LT，R _LDPC)，j≥2 (9)

The selection of degree j can influence two things.The first, it has determined LDPC to be coded in the performance of fading channel.The second, it controls R _LTRatio.Usually above-mentioned optimization problem is difficult to solve, because f ₆In R _LTBe the decreasing function of j, and R _LDPCIt is the increasing function of j.For finding preferred plan, with the exhaustive search algorithm with all R _LTAnd R _LDPCValue all list rice, again the selection wherein only value as design.

Simulation result

Supposing has 500 source information in each Frame the inside.Fig. 5 and Fig. 6 are given in R respectively _cFixing and at R _cThroughput performance during for variable.

Fig. 5, when K=500, the balance between physical layer LDPC coding and the network layer LT coding is with the variation of the setting of δ.Chain-dotted line is the optimal throughput performance of system.

Fig. 6, when K=500, the balance between physical layer LDPC coding and the network layer LT coding is with R _cThe variation of setting.Chain-dotted line is the optimal throughput performance of system.

Fig. 5 works as R _cIn the time of fixedly, adopt code check r _LDPC=1/2 (3,6) regular LDPC coding rice is encoded to the data segment of network layer.Reduce δ and can increase redundancy, can increase the success rate of decoding, thereby improve the throughput of whole system in network layer.Simulation result shows, when δ=0.001, the whole system throughput is maximum.

Fig. 6 is various j when spending, different R _cThe optimal throughput performance of value.In low SNR zone, because the code check of LDPC coding is very high, throughput of system is zero.Throughput increases Shi Caihui at SNR and increases.The degree j of systematic function quality and array code is related.At the zone line of SNR, the value of degree j is big more, and throughput performance is good more, and for example, when SNR=7dB, the throughput performance of j=8 is better than when j=4.Yet along with the value of SNR increases, the value of degree j is more little, and throughput performance is good more.For example, when SNR=10dB, the throughput performance of j=4 is good when higher than the j value.So want the value rice of careful this degree of selection j to be equilibrated at the setting of whole SNR.Can find out from simulation result, when SNR=10dB, j=6 (corresponding R _c=0.793) be a best selection, this moment, the throughput of system can reach 0.55.

Above-described appearance is a preferred implementation of the present invention, the invention is not restricted to above embodiment.Be appreciated that other improvement and variation that those skilled in the art directly derive or associate under the prerequisite that does not break away from basic design of the present invention, all should think to be included within protection scope of the present invention.

Claims (7)

1. the Radio Network System throughput optimization method based on no rate coding is characterized in that, comprises the steps:

1) single transmitter sends information to single or a plurality of receiving terminals, and information symbol is divided into Frame, and each Frame comprises K source symbol s={s ₁, s ₂..., s _K;

2) through the LT encoder said Frame is encoded, producing length continuously is L _cCoded data packet;

3) add header P for each coded data packet _hSymbol and CRC P _CRCSymbol forms length L _s=L _c+ P _h+ P _CRCCoded data packet;

4) pass through the LDPC encoder with R _cEncoding rate is L with said length _sThe coded data packet encoder, obtain length N _t=L _s/ R _cLinear block codes word C, said code word C carries out the radio channel transmission through BPSK modulation;

5) for each Frame, transmitter constantly sends coded data packet, receives from the affirmation information of all predetermined receiving terminals or the maximum coded data packet quantity that reached permission up to transmitter just to stop to send, and sends next Frame then;

6) carry out LDPC decoding at each receiving terminal, any code word that receives in GF (2) territory

does not satisfy condition

all will be considered to invalid and abandon whole bag;

7) detect carrying out CRC,, then abandon whole bag if can't pass this detection through the data of LDPC decoding;

8) will send to the LT decoder through the correct decoded data that CRC detects, carry out the LT decoding;

9) after receiving terminal is decoded successfully, send confirmation to transmitter.

2. according to the said Radio Network System throughput optimization method of claim 1, it is characterized in that: the throughput T of said Radio Network System based on no rate coding _sFor

$T_s=\frac{K}{N_t}\&CenterDot;R_{\mathrm{LDPC}}\&CenterDot;R_{\mathrm{LT}},$

R wherein _LDPCBe the correct Solution code rate of LDPC decoder, R _LTIt is the ratio that the information of LT decoder successfully receives.

3. according to the said Radio Network System throughput optimization method based on no rate coding of claim 1, it is characterized in that: in the network layer of Radio Network System, adopt powerful orphan's degree LT coding to distribute, the degree distribution function does D wherein _MaxBe maximum degree during powerful orphan's degree distributes, ρ _jBe percentage with check-node of j degree, ρ _j∈ (0,1);

Said degree distribution function is converted into

S wherein _MaxBe the degree of maximum variable node, s _iBe the ratio of the variable node of degree of having i, be expressed as

$s_i=\frac{e^{-\mathrm{\&lambda;}}{\mathrm{\&lambda;}}^i}{i!},$

The length L of encoded packets _cUnder the probability of 1-δ, satisfy L _c=KZ, wherein Z=∑ _dρ (d)+τ (d), ρ (d) and τ (d) are:

ρ(1)=1/K

$\mathrm{\&rho;}\left(d\right)=\frac{1}{d(d-1)},$ d=2，3，…，K

$\mathrm{\&tau;}\left(d\right)=\left\{\begin{array}{c}\frac{S}{K}\frac{1}{d},d=\mathrm{1,2},...(K/S)\\ \frac{S}{K}\log (S/\mathrm{\&delta;}),d=K/S\\ 0,d>K/S\\ \end{array}\right.$

Wherein,

C is a real number between 0 and 1; δ representes probability of decoding failure; Code word size behind network layer LT coding is L _c=K+2log _e(S/ δ) S.

4. according to the said Radio Network System throughput optimization method of claim 1 based on no rate coding; It is characterized in that: in the physical layer of Radio Network System, adopt the array code, the check matrix H of array code is by three parameter determining: prime number p and integer k and j; K, j≤p; K and j be in the H matrix in the row and column element be 1 number; H is the matrix of jp * kp, is expressed as:

Wherein I is p * p cell matrix, and α is the single left side of representative or the permutation matrix of right cyclic shift;

Code word size N behind the LDPC coding _tLength K with message part in the code word _LDPCBe respectively N _t=k * p and K _LDPC=(k-j) * p; Encoding rate is R _c=K _LDPC/ N _tThe code word size that generates in physical layer LDPC coding back is:

$N_t=\frac{(K/2{\log }_e(S/\mathrm{\&delta;})S+P_{\mathrm{CRC}}+P_h)}{R_c}.$

5. according to claim 3 or 4 said Radio Network System throughput optimization methods, it is characterized in that based on no rate coding:

Work as R _cIn the time of fixedly, the encoded packets length L _c=K+2Slog _e(S/ δ), wherein S is the monotonically decreasing function of δ, K is the length of source information piece, L _cBe the monotonically decreasing function of δ,

$L_c=K+2\left({\log }_e(K/\mathrm{\&delta;})\sqrt{K}\right){\log }_e\left(\frac{{\log }_e(K/\mathrm{\&delta;})\sqrt{K}}{\mathrm{\&delta;}}\right),$

L _cThe function that is expressed as δ is L _c=f ₁(δ), the throughput T of system _sDepend on K, N _t, R _LTAnd R _LDPC, N wherein _t=L _c/ R _c, R _LTAnd R _LDPCBe the ratio of every layer of successful decoded information, R _LTWith the relation function of δ be R _LT=α ^δ, a ∈ (0,1) wherein;

R _LDPCValue near 1, T then _s(δ)=f ₂(L _c, δ, a ^δ)

$T_s\&ap;\frac{{\mathrm{KR}}_c}{L_c}{\mathrm{\&alpha;}}^{\mathrm{\&delta;}},$

Being expressed as function is T _s(δ)=f ₂(L _c, δ, a ^δ), the maximum throughput of Radio Network System is maxT _s(δ)=f ₂(L _c, δ, α ^δ), δ>0;

Work as R _cWhen variable, physical layer code check R _cFor

$R_c=\frac{K_{\mathrm{LDPC}}}{N_t}=\frac{(k-j)\&times;p}{k\&times;p}=1-\frac{j}{k},$

Wherein, R _cBy j and k decision, R _LDPCBe expressed as R _LDPC=f ₃(j), R _LDPCIt is the dull increasing function of j; K _LDPCBe expressed as K _LDPC=f ₄(j), K _LDPCIt is the monotonically decreasing function of j; The ratio R of the correct decoding of the information of network layer _LTBe the monotonically decreasing function of j, be expressed as R _LT=f ₅(j); The maximum throughput of Radio Network System is maxT _s(j)=f ₆(K, R _LT, R _LDPC), j>=2.

6. according to the said Radio Network System throughput optimization method based on no rate coding of claim 5, it is characterized in that: when δ=0.001, the throughput of whole Radio Network System is maximum.

7. according to the said Radio Network System throughput optimization method of claim 5, it is characterized in that: work as SNR=10dB, j=6, corresponding R based on no rate coding _c=0.793 o'clock, the throughput of whole Radio Network System was maximum.

Images