CN102387002B - Method used for reducing wireless network frame loss rate based on adaptive coding modulation - Google Patents

Abstract

The invention relates to a method used for reducing wireless network frame loss rate based on adaptive coding modulation, which comprises the following steps: determining an initial value range of goal error rate; and calculating the length of the value range, and judging whether the value range is smaller than a pre-set value, if the value range is smaller than the pre-set value, the best goal error rate is (Pmax+Pmin)/2, otherwise, a testing point is calculated, and a golden section method is used for seeking the best goal error rate, so as to enable the frame loss rate to be minimum, namely the network throughput is maximum.

Description

A kind of for reducing the method based on adaptive coding and modulating wireless network frame loss rate

Technical field

The present invention relates to a kind of for reducing the method based on adaptive coding and modulating wireless network frame loss rate.

Background technology

In traditional network design scheme, for effectively utilizing Internet resources, improve network throughput, we generally adopt Adaptive Modulation (AMC) technology in physical layer.Adaptive Modulation is under different channel statuss, selects different transmission modes, to improve network throughput.In the choosing of transmission mode, most important parameter is exactly the selection of target error rate, under different target error rate, the network performance obtaining has very large difference, fixing target error rate is by the performance of limiting network to a certain extent, therefore select a good target error rate, can effectively improve network throughput.

Fig. 1 is the analogous diagram between existing packet loss and target error rate, by analyzing the performance of packet loss under the different target error rate, we can notice no matter the signal to noise ratio (SNR) of channel is much, packet loss becomes conic section (minimum point that solid five-pointed star is curve), i.e. ξ=ξ (P substantially with the variation of target error rate ₀) within the scope of whole target error rate, be unimodal function, and we cannot obtain its expression, more can not write its derivative expressions.Therefore being also difficult for finding out suitable minimum packet loss makes the throughput of network reach maximum.

Summary of the invention

For fear of above the deficiencies in the prior art, the invention provides a kind of for reducing the method based on adaptive coding and modulating wireless network frame loss rate.

Object of the present invention is achieved through the following technical solutions.

For reducing the method based on adaptive coding and modulating wireless network frame loss rate, the method comprises the steps:

1) determine the initial interval [P of target error rate _min, P _max];

2) calculate interval length, judge whether interval is less than predefined value, if be less than this value, the optimum target error rate is (P _max+ P _min)/2; Otherwise, calculate and sound out point:

λ＝P _min+0.382*(P _max-P _min)

μ＝P _min+0.618*(P _max-P _min)；

Calculate ξ (λ), ξ (μ) goes to step 3 when ξ (λ) > ξ (μ), when ξ (λ)≤ξ (μ), go to step 4, ξ (λ), it is λ that ξ (μ) is target error rate, the value of packet loss during μ;

3) put

$\left\{\begin{array}{c}{P}_{\min }=\mathrm{\λ}\\ P_{\max }=P_{\max }\\ \mathrm{\λ}=\mathrm{\μ}\\ \mathrm{\μ}=P_{\min }+0.618*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2;

4) put

$\left\{\begin{array}{c}{P}_{\min }=P_{\min }\\ P_{\max }=\mathrm{\μ}\\ \mathrm{\μ}=\mathrm{\λ}\\ \mathrm{\λ}=P_{\min }+0.382*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2.

Further, described predefined value is arbitrarily small accuracy value.

Further, described initial interval [P _min, P _max] be [0.0001,0.1].

The present invention can effectively utilize Internet resources, reduces network packet loss rate, improves network throughput.

Accompanying drawing explanation

Fig. 1: the analogous diagram between existing packet loss and target error rate;

Fig. 2: the inventive method flow chart;

Fig. 3: the packet loss that the inventive method obtains and the analogous diagram between target error rate;

Fig. 4: the packet loss that the inventive method obtains and the packet loss Performance Ratio under prior art are.

Embodiment

The present invention changes by investigating the throughput of different error rate lower network, proposes to find best target error rate by Fibonacci method, makes packet loss reach minimum, and network throughput reaches maximum.

First analyze the packet loss performance in Adaptive Modulation lower network below.

The target error rate P of general hypothesis physical layer ₀for certain value,, according to the target error rate requirement of supposition, there is a modulation system threshold vector γ=[γ who comprises N element ₀, γ ₁..., γ _n+1,] ^t, whole received signal to noise ratio region [0 ,+∞] is divided into the region (γ wherein of N+1 non-overlapping copies ₀=0, γ _n+1=∞).Adopt certain algorithm to determine modulation thresholding

select γ _nmake the average Packet Error Ratio under every kind of modulation system all equal target Packet Error Ratio P ₀: total like this Packet Error Ratio

must equal P ₀.The interval corresponding a kind of rate mode respectively of each signal to noise ratio.Its channel model adopts finite-state channel metastasis model (FSMC) to describe.

Data link layer adopts limit for length buffering area, and therefore, when buffering area queue is full, newly arrived packet will overflow, and supposes that bag flood rate is P _d, packet is only correctly transmitted in channel, and is just counted as merit while not overflowed and transmits, so packet loss can be expressed as:

ξ=1-(1-P _d) (1-P _ch), network throughput can be expressed as:

η=(λ T _f) (1-ξ)=(λ T _f) (1-P _d) (1-P ₀), λ T wherein _fthat bag reaches rate.Therefore can find out throughput and the packet loss performance of wanting critic network, topmost problem is calculation overflow bag rate P _d.

For analyzing the bag rate of overflowing of network, we carry out following Several Analysis.

1) first analysis node place obtains inlet flow situation.The inlet flow distribution obedience parameter of supposing node is λ T _fpoisson while distributing, it is distributed as:

$P({\mathrm{\Θ}}_t=a)=\frac{{\left({\mathrm{\λT}}_f\right)}^a{e}^{(-{\mathrm{\λT}}_f)}}{a!}---\left(1\right)$

2) queue service state is analyzed.The bandwidth of supposing channel is b, and the number-of-packet that every frame can transmit is: c _n=bR _n.

3) queue and service state steady-state distribution.The state-transition matrix of supposing associating queue length and service speed is

wherein matrix element is defined as:

$P_{(u,c)(v,d)}^E=P(C_{t+1}=d/C_t=c)\×P(U_t=v|U_{t-1}=u,C_t=c)---\left(2\right)$

Rate transitions probability P (C wherein _t+1=d/C _t=c) with FSMC model, calculate P (U _t=v/U _t-1=u, C _t=c) can further be decomposed into:

$P(U_t=v/U_{t-1}=u,C_t=c)$

$=\left\{\begin{array}{cc}P({\mathrm{\&Theta;}}_t=v-\max \{0,u-c\left\}\right)& \mathrm{if}0\&le;v<K\\ 1-\underset{0\&le;v<K}{\mathrm{\&Sigma;}}P(U_t=v|U_{t-1}=u,C_t=c)& \mathrm{ifv}=K\end{array}\right.---\left(3\right)$

Through above-mentioned analysis, can obtain due to the number average of the full packet overflowing in buffering area be:

$E\left\{D\right\}=\underset{\&ForAll;a,\&ForAll;u,\&ForAll;c}{\mathrm{\&Sigma;}}\max \{0,a-K+\max \{0,u-c\left\}\right\}---\left(4\right)$

$\&times;P(\mathrm{\&Theta;}=a)\&times;P(U=u,C=c)$

Wherein K is buffering area maximum queue length.Further obtaining datagram overflow rate is:

$P_d=\frac{E\left\{D\right\}}{E\left\{\mathrm{\&Theta;}\right\}}=\frac{E\left\{D\right\}}{{\mathrm{\&lambda;T}}_f}---\left(5\right)$

By above-mentioned derivation, we can find out, when we determine target error rate P ₀value time, just can calculate the bag rate P that overflows _dvalue, so packet loss ξ can be regarded as the function of target error rate,

ξ＝ξ(P ₀) (6)

The present invention sounds out point and carries out the comparison of functional value by getting, and the region of search that comprises minimal point is constantly shortened, and when interval shortens to a certain degree, the functional value of the each point on interval all approaches minimum, thereby each point can be seen the approximate of minimal point as.These class methods are called again dividing method, typically have Fibonacci method (0.618 method) and Fibonacci method (Fibonacci method).Therefore here we can adopt Fibonacci method to ask the optimum target error rate, make packet loss minimum.

Fibonacci method is also 0.618 method, it is the minimal point searching algorithm based on range shortening, by constantly dwindling the region of search, is here the span of target error rate, finally make the end points of the region of search, the value of target error rate is approached the smallest point of packet loss.

In the present invention, we suppose two end points of the region of search, i.e. the minimum value P of target error rate _min=0.0001, maximum occurrences P _max=0.1, first Fibonacci method generates two interior points, i.e. two of target error rate sensing point p according to golden ratio ₁and p ₂, wherein

p ₁＝P _min+0.382*(P _max-P _min)

p ₂＝P _min+0.618*(P _max-P _min)

Then according to ξ (p ₁), ξ (p ₂) magnitude relationship reselect the region of search.If ξ is (p ₁) < ξ (p ₂), the region of search becomes [P _min, P _max]; If ξ is (p ₁) > ξ (p ₂), the region of search becomes [P _min, p ₂];

Concrete algorithm steps is:

1. select [P between original area _min, P _max]=[0.0001,0.1], and arbitrarily small precision ε > 0

2. calculate P _max-P _minvalue, if P _max-P _min< ε, stops calculating.Otherwise, calculate and sound out point:

λ＝P _min+0.382*(P _max-P _min)

μ＝P _min+0.618*(P _max-P _min)；

And calculate ξ (λ) by formula (1)-(6), the value of ξ (μ) turns 3 when ξ (λ) > ξ (μ), when ξ (λ)≤ξ (μ), turns 4;

3. put

$\left\{\begin{array}{c}{P}_{\min }=\mathrm{\&lambda;}\\ P_{\max }=P_{\max }\\ \mathrm{\&lambda;}=\mathrm{\&mu;}\\ \mathrm{\&mu;}=P_{\min }+0.618*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2;

4. put

$\left\{\begin{array}{c}{P}_{\min }=P_{\min }\\ P_{\max }=\mathrm{\&mu;}\\ \mathrm{\&mu;}=\mathrm{\&lambda;}\\ \mathrm{\&lambda;}=P_{\min }+0.382*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2;

From above-mentioned algorithm, finally work as P _max-P _minduring < ε, circulation finishes, and the optimum target error rate can be expressed as (P _max+ P _min)/2, the minimum packet loss of trying to achieve is ξ ((P _max+ P _min)/2).

Above algorithm can be used the flowcharting of Fig. 2.

Utilize the Fibonacci method of above mentioning to be optimized the target error rate of network, find the minimum packet loss under different signal to noise ratios, as shown in Figure 3.Wherein in figure, solid circles is the little packet loss point of doing obtaining by Fibonacci method, and as can be seen from the figure, it overlaps with solid five-pointed star substantially.Be that we have found optimum target error rate value by Fibonacci method, make network packet loss rate minimum.

This method of the present invention can effectively reduce network packet loss rate, Fig. 4 utilizes network packet loss rate performance that Fibonacci method obtains and target error rate to be respectively 0.1,0.01,0.001 and the comparison of the packet loss performance of 0.0001 o'clock, as can be seen from Figure 4, the packet loss performance of the method is better than the situation that any target error rate is certain.

Claims (3)

1. for reducing the method based on adaptive coding and modulating wireless network frame loss rate, it is characterized in that, the method comprises the steps:

1) determine the initial interval [P of target error rate _min, P _max];

2) calculate interval length, judge whether interval is less than predefined value, if be less than this value, the optimum target error rate is (P _max+ P _min)/2; Otherwise, calculate and sound out point:

λ=P _min+0.382*(P _max-P _min)

μ=P _min+0.618*(P _max-P _min)；

Calculate ξ (λ), ξ (μ) goes to step 3 when ξ (λ) > ξ (μ), when ξ (λ)≤ξ (μ), go to step 4, ξ (λ), it is λ that ξ (μ) is target error rate, the value of packet loss during μ;

3) put

$\left\{\begin{array}{c}{P}_{\min }=\mathrm{\&lambda;}\\ P_{\max }=P_{\max }\\ \mathrm{\&lambda;}=\mathrm{\&mu;}\\ \mathrm{\&mu;}=P_{\min }+0.618*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2;

4) put

$\left\{\begin{array}{c}{P}_{\min }=P_{\min }\\ P_{\max }=\mathrm{\&mu;}\\ \mathrm{\&mu;}=\mathrm{\&lambda;}\\ \mathrm{\&lambda;}=P_{\min }+0.382*(P_{\max }-P_{\min })\end{array}\right.,$ Turn 2;

The concrete steps of described packet loss are:

1) first analysis node place obtains inlet flow situation, supposes that the inlet flow distribution obedience parameter of node is λ T _fpoisson while distributing, it is distributed as:

$P({\mathrm{\&Theta;}}_t=a)=\frac{{\left({\mathrm{\&lambda;T}}_f\right)}^a{e}^{(-\mathrm{\&lambda;}{T}_f)}}{a!}---\left(1\right)$

2) queue service state is analyzed, and the bandwidth of supposing channel is b, and the number-of-packet that every frame can transmit is: c _n=bR _n,

3) queue and service state steady-state distribution, suppose that the state-transition matrix of associating queue length and service speed is

wherein matrix element is defined as:

$P_{(u,c)(v,d)}^E=P(C_{t+1}=d/C_t=c)\&times;P(U_t=v|U_{t-1}=u,C_t=c)---\left(2\right)$

Rate transitions probability P (C wherein _t+1=d/C _t=c) with FSMC model, calculate P (U _t=v/U _t-1=u, C _t=c) can further be decomposed into:

$\begin{array}{c}P(U_t=v/U_{t-1}=u,C_t=c)\\ =\left\{\begin{array}{cc}P({\mathrm{\&Theta;}}_t=v-\max \{0,u-c\left\}\right)& \mathrm{if}0\&le;v<K\\ 1-\underset{0\&le;v<K}{\mathrm{\&Sigma;}}P(U_t=v|U_{t-1}=u,C_t=c)& \mathrm{ifv}=k\end{array}\right.\end{array}---\left(3\right)$

Through above-mentioned analysis, can obtain due to the number average of the full packet overflowing in buffering area be:

$\begin{array}{c}E\left\{D\right\}=\underset{\&ForAll;a,\&ForAll;u,\&ForAll;c}{\mathrm{\&Sigma;}}\max \{0,a-K+\max \{0,u-c\left\}\right\}\\ \&times;P(\mathrm{\&Theta;}=a)\&times;P(U=u,C=c)\end{array}---\left(4\right)$

Wherein K is buffering area maximum queue length, further obtains datagram overflow rate to be:

$P_d=\frac{E\left\{D\right\}}{E\left\{\mathrm{\&Theta;}\right\}}=\frac{E\left\{D\right\}}{\mathrm{\&lambda;}{T}_f}---\left(5\right)$

As definite target error rate P ₀value time, just calculate the bag rate P that overflows _dvalue, so packet loss ξ can be regarded as the function of target error rate,

ξ=ξ(P ₀) （6）。

2. according to claim 1 a kind ofly it is characterized in that for reducing the method based on adaptive coding and modulating wireless network frame loss rate, described predefined value is arbitrarily small accuracy value.

3. according to claim 1 a kind ofly it is characterized in that for reducing the method based on adaptive coding and modulating wireless network frame loss rate, described initial interval [P _min, P _max] be [0.0001,0.1].

Images