CN102970170B - A kind of establishing method of integer weight of network transmission quantity - Google Patents

Abstract

The invention discloses a kind of establishing method of integer weight of network transmission quantity, belong to data weights setting technique field.The steps include: that (1) obtains the transmission volume p of m platform data transmission unit by hardware monitoring _i(i=1 ..., m); (2) territory algorithm is sought: on the basis retaining m platform data transmission unit transmission volume numerical characteristic, find integer region [0, xm]; (3) Q value weighs algorithm surely: utilize Q value in integer region [0, xm], select suitable weights to distribute to m platform data transmission unit, thus calculate the weights of m platform data transmission unit.The weight values that the present invention effectively reduces data transmission unit is interval, thus integer weight of network transmission quantity and the such as physical concept such as wrap count, interval number of seconds can be closely linked to use flexibly as optimization of network parameters, without the need to secondary calculating, avoiding traditional weights method to set up cannot the deficiency of analog network transmission environment truly, improves network optimization efficiency.

Description

A kind of establishing method of integer weight of network transmission quantity

Technical field

The present invention relates to data weights setting technique field, more particularly, relate to a kind of establishing method of integer weight of network transmission quantity.

Background technology

Before elaboration technology contents of the present invention, now first the technical term of this area is done a relevant explanation, specifically as shown in table 1.

The basic conception of table 1 art technology noun is explained

Along with the fast development of the Internet, computer network has circulated the every aspect of human lives, and amount of information presents explosive growth.In order to allow the more effective running of network, people create a lot of network optimized approach, make network performance reach the optimal balance point of our needs by various hardware or software engineering.Transmission volume, as the important parameter of in network environment, is usually monitored by emphasis.Transmission quantity Network Based is that each data transmission unit arranges weights, and then carries out the network optimization according to these weights, often can play a multiplier effect.So, how just to have become an important topic according to transmission volume for each data transmission unit rationally arranges weights.

Based on above background, the weights of data transmission unit arrange below demand fulfillment 3 requirement: the numerical characteristic of (1) energy representative data transmitting element; (2) integer; (3) numerical intervals is as far as possible little.

Traditional weights method to set up mainly contains following three kinds: (1) for after least unit by unified for the transmission volume of each data transmission unit, directly it can be used as weights, is called original weights; (2) to the transmission volume of each data transmission unit by sorting from small to large, then use integer sequence (1,2,3 ...) be numbered as weights, be called sequence number weights; (3) transmission volume of each data transmission unit is added up, then calculate each data transmission unit and account for the percentage of total amount and as its weights after simple integer, be called percentage weights.

The shortcoming of above-mentioned traditional weights method to set up:

(1) original weights are the simplest weights, because it does not almost have any data transformations, and meet weights the numerical characteristic of transmitting element " can representative data " and " integer " two requirements in requirement be set, but the numerical intervals of transmission volume generally crosses over B, KB, MB tri-data organizational levels, unit is unified for after B, the numerical difference of each data transmission unit transmission volume is apart from often very large, almost directly can not be used for optimized network, because the parameter major part relevant to the network optimization is all little number field scope, such as the number of transmissions, transmission intercal, the network bandwidth etc., so secondary conversion must be carried out to original weights,

(2) sequence number weights meet weights and arrange " integer " and " numerical intervals is as far as possible little " two requirements in requirement, but it only remains the size sequencing information of each data transmission unit transmission volume, and do not reflect the numerical characteristic of each transmission volume particularly, thus modest efficacy, can not analog network transmission environment truly;

(3) percentage weights meet weights and arrange the numerical characteristic of transmitting element " can representative data " and " integer " two requirements in requirement, and the use of majority of network transmission situation can be met, particularly when the numerical difference distance of each data transmission unit transmission volume is larger, good effect can be played, but its numerical intervals is fixed on [0,100], for each data transmission unit numerical difference apart from less situation, will be seemed numerical intervals redundancy, if be now directly used for optimized network environment, the efficiency of the network optimization will be affected.

Summary of the invention

1. invent the technical problem that will solve

The object of the invention is to overcome in prior art transmission volume weights method to set up cannot analog network transmission environment truly, the deficiency of network optimization efficiency is affected with traditional weights method to set up, provide a kind of establishing method of integer weight of network transmission quantity, the present invention is mapped in the suitable integer region of scope on the basis retaining the original numerical characteristic of transmission volume data, thus can analog network transmission environment more truly, improve the efficiency of the network optimization, is the establishing method of the integer weights of a set of scientific and effective transmission volume.

2. technical scheme

For achieving the above object, technical scheme provided by the invention is:

In the present invention, basic conception is explained as shown in table 2.

Table 2 basic conception of the present invention is explained

The establishing method of a kind of integer weight of network transmission quantity provided by the present invention, the steps include:

(1) the transmission volume p of m platform data transmission unit is obtained by hardware monitoring _i;

(2) seek territory algorithm: on the basis retaining m platform data transmission unit transmission volume numerical characteristic, find integer region [0, xm], wherein expect that weights x is the weights mathematic expectaion of m platform data transmission unit;

(3) Q value weighs algorithm surely: utilize Q value in integer region [0, xm], select suitable weights to distribute to m platform data transmission unit, thus calculate the weights of m platform data transmission unit.

As the present invention further, the computational process of seeking territory algorithm in step (2) is as follows:

1) the upper dividing value of computing unit weights diversity factor for m platform expects that weights are unit weights diversity factor maximum in all weights integrated modes of the data transmission unit of x, under general situation, the unit weights diversity factor of this integrated mode is maximum, so obtain computing formula as follows:

$\max \left(A_m^x\right)=\frac{1}{x}\sqrt{\frac{{(m-1)(x-1)}^2+{\left[(x-1)(m-1)\right]}^2}{m}}---\left(c\right)$

Wherein: m is data transmission unit number of units, x is for expecting weights; Calculate according to above-mentioned formula (c)

2) computing unit transmission diversity factor T _m, formula T _mas follows:

$T_m=\frac{\sqrt{\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{(p_i-\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{p}_j)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{p}_j}---\left(b\right)$

Wherein: p _ifor the transmission volume of data transmission unit i; for the averaging network transmission quantity of each data transmission unit; M is data transmission unit number of units;

3) according to the number of units m of data transmission unit with expect weights x, expect that weights x is from little, query steps 1) dividing value in the unit weights diversity factor that calculates find first and be greater than step 2) unit that calculates transmission diversity factor T _munit weights diversity factor on dividing value get corresponding x as expectation weights, obtaining total weight value number is xm, the integer region [0, xm] that the scope that namely have found is suitable.

The theoretical foundation of seeking territory algorithm in the present invention is as follows:

1) diversity factor is expected with unit weights for standard:

Unit weights expect diversity factor represent m platform as one and expect that weights are the aggregate feature of the data transmission unit weights distribution of x, its meaning is to represent that m platform expects that weights are the mathematical expectation of the weights difference degree of data transmission unit under various weighed combination pattern of x.Because unit weights expect diversity factor the mean difference degree of all weights integrated modes can be reflected, so it more can represent m platform expect that weights are the weights difference degree of the data transmission unit of x, by studying the Changing Pattern of weights difference degree along with x and m that it can find data transmission unit.

$A_m^x=\frac{1}{x}\sqrt{\frac{\underset{k=1}{\overset{C_{\mathrm{xm}-1}^{m-1}}{\mathrm{\Σ}}}\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{(x_{\mathrm{ki}}-x)}^2}{C_{\mathrm{xm}-1}^{m-1}}}---\left(e\right)$

Variable declaration in table 3 formula (e)

Calculate m=2,3,4,5,6,7,8, x=2 ..., the unit weights of 20 expect diversity factor show as shown in table 4.

Table 4 unit weights expect diversity factor table

Following tendency chart can be drawn, as shown in Figure 2 from upper table.Can find from Fig. 2: (1), when expecting that weights x is certain, along with the increase of m number, unit weights expect diversity factor increase; (2) when data transmission unit number of units m is certain, along with the increase expecting weights x, unit weights expect diversity factor increase, and growth trend slows down gradually.

Unit transmission diversity factor T _mdiversity factor is expected with unit weights more close, namely absolute value is less, then illustrate that the weights of data transmission unit arrange the possibility that there is high similarity between transmission volume larger.So for given m, the weights method to set up that a kind of expectation weights x corresponding with data transmission unit transmission volume is minimum necessarily can be found.

Unit weights for such as following table 5 expect diversity factor table.

Table 5 unit weights expect diversity factor table

First corresponding data transmitting element number of units m column is found, then by order comparing unit transmission diversity factor T from top to bottom _mdiversity factor is expected with unit weights find and make absolute value minimum get corresponding x value as expecting weights.

Computing unit weights expect diversity factor java program ExceptCOV.java mainly as follows:

2) with dividing value in unit weights diversity factor for standard:

Diversity factor is expected from unit weights computing formula can find out, its amount of calculation can increase in geometry level along with the increase of m and x, and result accuracy also can reduce thereupon, works as m=7, and during x=16, result of calculation is just beyond program computation scope.So select during practical application with dividing value in unit weights diversity factor unit weights are replaced to expect diversity factor

Dividing value in unit weights diversity factor as the numerical characteristic of a data transmission unit weights distribution, it can reflect that m platform expects that weights are maximum weights diversity factor under all weights integrated modes of the data transmission unit of x, if be also T _mbe less than then necessarily have and a kind ofly expect that weights are the transmission volume that the weighed combination pattern of x is suitable for each data transmission unit, from computing formula (c) can find out, its calculating is fast and convenient, and amount of calculation can't increase along with the increase of x and m.Formula (c) is utilized to calculate m=2 ..., 7, x=2 ..., the result of 100 as shown in Figure 3.

Can find from Fig. 3, dividing value in unit weights diversity factor numeric distribution and unit weights expect diversity factor basically identical, for any one m, dividing value in unit weights diversity factor substantially x all tends to be steady when x=100, so can meet the demand of practical application substantially in [2,100] interval.

For dividing value in such as following table 6 unit weights diversity factor table.

Dividing value in table 6 unit weights diversity factor table

As shown in table 6, first find m column, then by order comparing unit transmission diversity factor T from top to bottom _mwith dividing value in unit weights diversity factor first is found to be greater than T _m's just get corresponding x value as expecting weights, obtaining total weight value number is xm, also namely have found suitable integer region [0, xm] and represents original

Originally seek territory algorithm to introduce in process and mainly employ following 4 computing formula, be unitedly described as follows:

(1) unit weights diversity factor A _m:

Unit weights diversity factor A _mthe numerical characteristic of m platform data transmission unit weights distribution is represented, the weights difference degree between main reflection m platform data transmission unit as one.

$A_m=\frac{\sqrt{\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{(x_i-\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{x}_j)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{x}_j}---\left(a\right).$

Variable declaration in table 7 formula (a)

(2) unit transmission diversity factor T _m:

Unit transmission diversity factor T _m, as the aggregate feature of the transmission volume of all data transmission unit, its meaning is to show m platform data transmission unit, the transmission difference degree of each data transmission unit.

$T_m=\frac{\sqrt{\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\Σ}}}{(p_i-\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{p}_j)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\Σ}}}{p}_j}---\left(b\right).$

Variable declaration in table 8 formula (b)

(3) the upper dividing value of unit weights diversity factor

Dividing value in unit weights diversity factor be exactly that m platform expects that weights are unit weights diversity factor maximum in all weights integrated modes of the data transmission unit of x.Easily find out and work as m=4, when expecting weight x=2, the unit weights diversity factor that (1,1,1,5) combine is maximum.In the same way, under general situation can be proved, the unit weights diversity factor of this integrated mode is maximum, so obtain computing formula, wherein: m is data transmission unit number of units, x for expect weights:

$\max \left(A_m^x\right)=\frac{1}{x}\sqrt{\frac{{(m-1)(x-1)}^2+{\left[(x-1)(m-1)\right]}^2}{m}}---\left(c\right).$

(4) similarity S _m:

Similarity represents the similarity degree that weights distribution distributes with transmission volume, the validation criteria that it can be arranged as weights, and its value is more close to 1, and illustrate that weights distribute and distribute more similar to transmission volume, also namely weights distribution is more reasonable.

$S_m=1-\frac{|T_m-A_m|}{T_m+A_m+1}---\left(d\right).$

As the present invention further, in step (3) Q value surely to weigh the computational process of algorithm as follows:

1) transmission volume of m platform data transmission unit is pressed the ratio of xm calculates;

2) m platform data transmission unit result step 1) calculated rounds;

3) deduct step 2 with total weight value number xm) m platform data transmission unit result sum, tentatively remained weights number k, often obtain new weight results in subsequent step and all can refresh k value, if the distribution of k value is over, then obtained the final weights of m platform data transmission unit, algorithm terminates;

4) just 1 part of weights is distributed to this data transmission unit if there is 10 in current weight result, k becomes k-1 and returns step 3); If having multiple is 0, then 1 part of weights being distributed to weight results is the data transmission unit that in the data transmission unit of 0, transmission volume is maximum, and k becomes k-1 and returns step 3); If there is no 0, carry out next step;

5) the Q value of m platform data transmission unit is calculated i=1,2 ..., m, wherein: n _ifor the weights that each data transmission unit has been distributed, p _ifor the transmission volume of data transmission unit i;

6) 1 part in residue weights is distributed to a maximum data transmission unit of Q value, k becomes k-1 and returns step 3).

In the present invention, Q value weighs the Q value derivation of equation of algorithm surely, is described as follows:

The situation of m=2:

First the situation of A, B two data transmission unit fair allocat weights is discussed, if the transmission volume of two data transmission unit is respectively p ₁and p ₂, the weights of existing distribution are respectively n ₁and n ₂, and n ₁and n ₂be not 0, then the mapping multiple of two sides is respectively p ₁/ n ₁and p ₂/ n ₂.Obviously only p is worked as ₁/ n ₁=p ₂/ n ₂time, it is fair and reasonable that mapping distribution is only, but usual p ₁/ n ₁≠ p ₂/ n ₂, and p _i/ n _ithe side that (i=1,2) numerical value is larger stands to lose.

Set up quantitative index to weigh inequitable degree:

If p ₁/ n ₁> p ₂/ n ₂, then to the relative inequality degree of A $r_A(n_1,n_2)=\frac{p_1/n_1-p_2/n_2}{p_2/n_2}---\left(1\right)$

If p ₂/ n ₂> p ₁/ n ₁, then to the relative inequality degree of B $r_B(n_1,n_2)=\frac{p_2/n_2-p_1/n_1}{p_1/n_1}---\left(2\right)$

Establish the quantitative index r weighing inequitable degree _a, r _bafter, the principle of specifying weights to distribute makes them little as far as possible exactly.

Determine allocative decision:

Utilize relative inequality degree r _aand r _b1 part of weights is discussed and distributes to A or B; P can be established without loss of generality ₁/ n ₁> p ₂/ n ₂, namely unfair to A, when distribution 1 part of weights, about p _i/ n _ithe inequality of (i=1,2) may have following 3 kinds of situations:

1) p ₁/ (n ₁+ 1) > p ₂/ n ₂even if this illustrates that A side increases by 1 weights, still unfair to A, so this 1 share should give A.

2) p ₁/ (n ₁+ 1) < p ₂/ n ₂, illustrate and will become B unfairness when A increases by 1 weights, reference (2) formula calculates and to the relative inequality degree of B is now

$r_B(n_1+1,n_2)=\frac{p_2(n_1+1)}{p_1{n}_2}-1---\left(3\right)$

3) p ₁/ n ₁> p ₂/ (n ₂+ 1), namely will become A unfair when B increases by 1 weights, reference (1) formula calculates and to the relative inequality degree of A is now

$r_A(n_1,n_2+1)=\frac{p_1(n_2+1)}{p_2{n}_1}-1---\left(4\right)$

Because the principle of fair allocat weights makes unjust Pingdu little as far as possible, if so

r _B(n ₁+1，n ₂)＜r _A(n ₁,n ₂+1)（5）

Then these 1 part of weights last should distribute to A; Otherwise then distribute to B, according to (3), (4) two formulas, (5) formula is equivalent to

$\frac{p_2^2}{n_2(n_2+1)}<\frac{p_1^2}{n_1(n_1+1)}---\left(6\right)$

Also it be easy to show that, the p of the first situation above-mentioned ₁/ (n ₁+ 1) > p ₂/ n ₂, also can cause (6) formula.So can reach a conclusion: when (6) formula is set up, last 1 part of weights should distribute to A, otherwise then distribute to B; Or, if note i=1,2 ..., m, then these 1 part of weights should distribute to the larger side of Q value.

Promote:

Said method can be generalized to the situation of m data transmitting element, if the transmission volume of i-th data transmission unit is p _i, distribute n _iindividual weights, (n _i≠ 0), i=1,2 ..., m, calculates:

$Q_i=\frac{p_i^2}{n_i(n_i+1)},$ i＝1,2,…,m（7）

These 1 part of weights is distributed to the maximum side of Q value, if residue weights also do not distribute, just by the weights n of now each data transmission unit _ias the starting point calculating Q value next time.

3. beneficial effect

Adopt technical scheme provided by the invention, compared with existing known technology, there is following remarkable result:

(1) under reality, the numerical intervals of transmission volume is very large, suitable numerical intervals must be found to convert to it, if numerical intervals is too large, the parameter of the network optimization directly can not be it can be used as, need secondary conversion, if namely the too little accuracy that can affect again calculating of numerical intervals also cannot go out Internet Transmission environment by real simulation, thus likely brings the error that a series of subsequent network is optimized.All the time, existing known technology all could not address this problem well, and the present invention adopts seeks territory algorithm and just can determine suitable integer region, tried one's best in the final integer region selected little and have enough spaces can show the difference degree of original each data transmission unit transmission volume, thus integer weight of network transmission quantity and the such as physical concept such as wrap count, interval number of seconds can be closely linked to use flexibly as optimization of network parameters, without the need to secondary calculating, improve network optimization efficiency;

(2) in addition, the Q value that the present invention adopts weighs algorithm surely, and ensure that similarity higher between weights and initial data, similarity reaches more than 0.980, therefore avoiding traditional weights method to set up cannot the deficiency of analog network transmission environment truly, decreases the error that subsequent network is optimized.When multiple stage data transmission unit transmits data simultaneously, the weights that the application of the invention is arranged distribute the limited network bandwidth, more sufficiently and reasonably can utilize network bandwidth resources; When multiple stage data transmission unit transmits data in turn, the weights that the application of the invention is arranged arrange the order of the transmission of data transmission unit in network data, can ensure that network application system can the change of corresponding transmitted data amount efficiently.

Accompanying drawing explanation

Fig. 1 is the flow chart of the establishing method of a kind of integer weight of network transmission quantity of the present invention;

Fig. 2 is that unit weights expect diversity factor tendency chart;

Fig. 3 is dividing value in unit weights diversity factor tendency chart;

Fig. 4 is m=4 network environment schematic diagram in embodiment 1;

Dividing value in unit weights diversity factor when Fig. 5 is m=4 in embodiment 1 tendency chart;

Fig. 6 is m=8 network environment schematic diagram in embodiment 2;

Dividing value in unit weights diversity factor when Fig. 7 is m=8 in embodiment 2 tendency chart.

Embodiment

For understanding content of the present invention further, the present invention is described in detail in conjunction with the accompanying drawings and embodiments.

Embodiment 1

The present embodiment is that 4 data transmission unit carry out Internet Transmission, the complete procedure of the establishing method of integer weight of network transmission quantity is shown by embodiment 1, its specific algorithm flow process as shown in Figure 1, is monitoring the transmission volume p of acquisition 4 data transmission unit by hardware _iafter, territory algorithm is sought in employing and Q value weighs algorithm surely, first finding a suitable integer region by seeking territory algorithm, in this integer region, then utilizing Q value surely to weigh the weights that algorithm calculates each data transmission unit, finally calculate similarity to verify that weights arrange effect.Its concrete steps are as follows:

(1) by the transmission volume p of hardware monitoring acquisition 4 data transmission unit _i, as shown in Figure 4, table 9 is each data transmission unit transmission volume statistics in fixed time section.

Each data transmission unit transmission volume statistics in table 9 fixed time section

Data transmission unit	Each data transmission unit transmission volume p in fixed time section i(unit B)
		1	9891
2	1000
		3	989
4	89

(2) territory algorithm is sought:

Algorithm object: the object of this algorithm is on the basis retaining each data transmission unit transmission volume numerical characteristic, finds a suitable integer region and sets weights.Selected cell transmission diversity factor of the present invention is as the numerical characteristic of data transmission unit transmission volume, so-called " suitable integer region " refers to that the final integer region selected has enough spaces can show the difference degree of original each data transmission unit transmission volume, and expects that weights x is also minimum.

The computational process of seeking territory algorithm is as follows:

1) the upper dividing value of computing unit weights diversity factor for m platform expects that weights are unit weights diversity factor maximum in all weights integrated modes of the data transmission unit of x, under general situation, the unit weights diversity factor of this integrated mode is maximum, so obtain computing formula as follows:

$\max \left(A_m^x\right)=\frac{1}{x}\sqrt{\frac{(m-1){(x-1)}^2+{\left[(x-1)(m-1)\right]}^2}{m}}---\left(c\right)$

Wherein: m is data transmission unit number of units, x is for expecting weights; Calculate according to above-mentioned formula (c)

When calculating m=4, x is from dividing value the unit weights diversity factor of 2 to 100 as shown in Figure 5.Observe Fig. 5 can find, the dividing value in unit weights diversity factor as x=20 substantially tend towards stability, also namely the span of x is just enough 2 to 19.Intercept x=2 ..., dividing value in the unit weights diversity factor of 19 as table 10.

Table 10x=2 ..., dividing value in the unit weights diversity factor of 19

2) computing unit transmission diversity factor T _m, formula T _mas follows:

$T_m=\frac{\sqrt{\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{(p_i-\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{p}_j)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{p}_j}---\left(b\right)$

Wherein: p _ifor the transmission volume of data transmission unit i; for the averaging network transmission quantity of each data transmission unit; M is data transmission unit number of units;

Computing unit transmission diversity factor in the present embodiment:

$T_4=\frac{\sqrt{\frac{1}{4}\&times;[{(9891-2992.25)}^2+{(1000-2992.25)}^2+{(989-2992.25)}^2+{(89-2992.25)}^2]}}{2992.25}\&ap;1.337.$

3) according to the number of units m of data transmission unit with expect weights x, expect weights be x from little, query steps 1) dividing value in the unit weights diversity factor that calculates table, finds first and is greater than step 2) unit that calculates transmits diversity factor T _munit weights diversity factor on dividing value get corresponding x as expectation weights, obtaining total weight value number is xm, the integer region [0, xm] that the scope that namely have found is suitable.By dividing value in query unit weights diversity factor in the present embodiment table finds, as x=5, and dividing value in unit weights diversity factor just greater than unit transmission diversity factor, so data transmission unit transmission volume summation 11969 is changed into each data transmission unit weights summation xm=5 × 4=20, also namely have found the suitable integer region of scope [0,20].

Dividing value in computing unit weights diversity factor in the present embodiment java program MaxCOV.java mainly as follows:

(3) Q value weighs algorithm surely: utilize Q value in integer region [0, xm], select suitable weights to distribute to m platform data transmission unit, thus calculate the weights of m platform data transmission unit.

Algorithm object: seek territory algorithm and have found suitable integer region [0, xm], and the object that Q value weighs algorithm surely utilizes a kind of standard and Q value in integer region [0, xm], select suitable weights to distribute to each data transmission unit exactly.So-called " suitable weights " refer to that the transmission volume similarity of weights and the original each data transmission unit finally distributing to each data transmission unit is high, and namely the ratio of each weights and transmission volume is consistent and weights summation is xm substantially.

The computational process that Q value weighs algorithm is surely as follows:

1) transmission volume of m platform data transmission unit is pressed the ratio of xm calculates.In the present embodiment, the ratio of the transmission volume of each data transmission unit according to 11969:20 is calculated, obtain non-integer weights n' _i(i=1,2,3,4) are as shown in table 11 below;

Table 11 each data transmission unit non-integer weights n' _i

Data transmission unit	Non-integer weights n' i
		1	16.53
2	1.67
		3	1.65
4	0.15

2) m platform data transmission unit result step 1) calculated tentatively rounds.In the present embodiment, above data are tentatively rounded and obtain following data, as shown in table 12:

Table 12 embodiment 1 tentatively round weights

Data transmission unit	Round weights n i
		1	16
2	1
		3	1
4	0

3) deduct step 2 with total weight value number xm) m platform data transmission unit result sum, obtain residue weights number k., distribute weights 16+1+1+0=18 in the present embodiment, so the weights that also residue 2 is to be allocated, i.e. k=2, often obtains new weight results and can refresh k value in subsequent step, if the distribution of k value is over, then obtained the final weights of m platform data transmission unit, algorithm terminates;

4) method of distributing residue weights is as follows: just 1 part of weights is distributed to this data transmission unit if there is 10 in current weight result, k becomes k-1 and returns step 3); If having multiple is 0, then 1 part of weights is distributed to the maximum data transmission unit of transmission volume, k becomes k-1 and returns step 3); If there is no 0, carry out next step;

Data transmission unit 4 is only had to be 0 in the current weight result of the present embodiment, so (just 1 part of weights is distributed to this data transmission unit if there is 10 in current weight result) on principle 1 part of weights should be distributed to it, k becomes 1, so each data transmission unit weights n when obtaining k=1 _i(i=1,2,3,4), as following table 13:

Each data transmission unit weights n during table 13k=1 _i

Data transmission unit	Each data transmission unit integer weights n during k=1 i
		1	16
2	1
		3	1
4	1

5) the Q value of m platform data transmission unit is calculated i=1,2 ..., m, wherein: n _ifor the weights that each data transmission unit has been distributed, p _ifor the transmission volume of data transmission unit i.In residue weights 1 part is distributed to a maximum data transmission unit of Q value, and k becomes k-1 and returns step 3).

(7) formula of Q value-based algorithm is utilized in the present embodiment i=1,2 ..., m calculates the Q value of each data transmission unit as following table 14.

The each data transmission unit Q value of table 14

Data transmission unit	Q value
		1	359676.03
2	500000
		3	489060.5
4	3960.5

Now the Q value of data transmission unit 2 is maximum, so 1 part in residue should be distributed to data transmission unit 2, k to become 0, so each data transmission unit final integer weights n when obtaining k=0 _i(i=1,2,3,4), as following table 15.

Each data transmission unit final integer weights n during table 15k=0 _i

Data transmission unit	Each data transmission unit final integer weights n during k=0 i
		1	16
2	2
		3	1
4	1

In the present embodiment, surely to weigh algorithm calculation procedure Qvalue.java mainly as follows for Q value:

After completing the setting of integer weight of network transmission quantity, carry out verification msg, detailed process is as follows:

(1) computing unit weights diversity factor A _m, represent the numerical characteristic of m platform data transmission unit weights distribution, the weights difference degree between main reflection m platform data transmission unit;

$A_m=\frac{\sqrt{\frac{1}{m}\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{(x_i-\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_j)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_j}---\left(a\right)$

Wherein x _ifor the weights of data transmission unit i, for the average weight of each data transmission unit, m is data transmission unit number of units;

Particularly, the unit weights diversity factor A of each data transmission unit is calculated in the present embodiment ₄for:

$A_4=\frac{\sqrt{\frac{1}{4}\&times;[{(16-5)}^2+{(2-5)}^2+{(1-5)}^2+{(1-5)}^2]}}{5}\&ap;1.273$

(2) similarity S is calculated _m, similarity represents the similarity degree that weights distribute and transmission volume distributes, the validation criteria that it can be arranged as weights, and its value is more close to 1, then illustrate that weights distribute and distribute more similar to transmission volume, also namely weights distribution is more reasonable;

$S_m=1-\frac{|T_m-A_m|}{T_m+A_m+1}---\left(d\right).$

Particularly, similarity S is calculated in the present embodiment ₄for:

$S_4=1-\frac{|A_4-T_4|}{A_4+T_4+1}=1-\frac{|1.273-1.337|}{1.273+1.337+1}\&ap;0.982.$

Embodiment is known thus: method of the present invention effectively reduces numerical intervals and ensure that higher similarity, and therefore avoiding traditional weights method to set up cannot the deficiency of analog network transmission environment truly.

In addition, the time complexity easily extrapolating algorithm of the present invention from the code logic of calculation procedure is linear complexity, the scale calculated using m as algorithm, and concrete reckoning process is as shown in table 16 below.

The time complexity of table 16 algorithm of the present invention

Algorithm	Time complexity
		Seek territory algorithm	O(m)
Q value weighs algorithm surely	O(m)
		Integer weights arrange algorithm	O(m)=O(m)+O(m)

The time complexity of algorithm is the important indicator of measure algorithm efficiency, the time complexity of algorithm of the present invention is linear time complexity O (m), be only second to constant time complexity O (1) and logarithmic time complexity O (logm), therefore efficiency of algorithm of the present invention is still very high, substantially can ensure without time delay the transmission volume of each data transmission unit that monitors according to hardware calculate weights, the change of transmitted data amount in quick response to network application system, thus ensure that the real-time of the network optimization, and then effectively reduce network failure rates.

Embodiment 2

The present embodiment is the overall process that 8 data transmission unit carry out Internet Transmission, and its concrete steps are as follows:

(1) by the transmission volume p of hardware monitoring acquisition 8 data transmission unit _i, as shown in Figure 6, table 17 is each data transmission unit transmission volume statistics in fixed time section.

Each data transmission unit transmission quantity statistics in table 17 fixed time section

Data transmission unit	Each data transmission unit transmission quantity p in fixed time section i(unit B)
		1	9891
2	1989
		3	989
4	898
		5	89
6	98
		7	9
8	8

(2) territory algorithm is sought:

The computational process of seeking territory algorithm is as follows:

1) calculate m=8 time, x from dividing value the unit weights diversity factor of 2 to 100, as Fig. 7.Observe Fig. 7 can find, the dividing value in unit weights diversity factor as x=20 substantially tend towards stability, also namely the span of x is just enough 2 to 19.Intercept x=2 ..., dividing value in the unit weights diversity factor of 19 as following table 18.

Table 18x=2 ..., dividing value in the unit weights diversity factor of 19

2) computing unit transmission diversity factor T _mas follows:

Computing unit transmission diversity factor:

$T_8=\frac{\sqrt{\begin{array}{c}\frac{1}{8}\&times;[{(9891-1746.375)}^2+{(1989-1746.375)}^2+{(989-1746.375)}^2+{(898-1746.375)}^2\\ +{(89-1746.375)}^2+{(98-1746.375)}^2+{(8-1746.375)}^2+{(9-1746.375)}^2]\end{array}}}{1746.375}\&ap;1.802.$

3) by finding in question blank 17, as x=4, dividing value in unit weights diversity factor just greater than unit transmission diversity factor, so data transmission unit transmission volume summation 13971 is changed into each data transmission unit weights summation

Namely xm=4 × 8=32, also have found the suitable integer region of scope [0,32].

(3) Q value weighs algorithm surely:

The computational process that Q value weighs algorithm is surely as follows:

1) ratio of the transmission quantity of each data transmission unit according to 13971:32 is calculated, obtain non-integer weights n' _i(i=1,2,3,4,5,6,7,8) are as following table 19.

Table 19 each data transmission unit non-integer weights n ' _i

Data transmission unit	Non-integer weights n ' i
		1	22.65
2	4.55
		3	2.27
4	2.06
		5	0.21
6	0.22
		7	0.02
8	0.02

2) above data are tentatively rounded obtain following data, shown in table 20.

Table 20 embodiment 2 round weights

Data transmission unit	Round weights n i
		1	22
2	4

3	2
		4	2
5	0
		6	0
7	0
		8	0

3) distribute weights 22+4+2+2=30, the weights that also residue 2 is to be allocated, i.e. k=2, we find existence 4 n in addition _i=0 (i=5,6,7,8) data transmission unit, uses the distribution principle (if having multiple in current weight result is 0, then 1 part of weights being distributed to the maximum data transmission unit of wherein transmission volume) of special circumstances, table 21 is data transmission unit 5 in fixed time section, 6,7, the transmission volume of 8.

Data transmission unit 5,6 in table 21 fixed time section, the transmission volume of 7,8

Data transmission unit	Each data transmission unit transmission volume p in fixed time section i(unit B)
		5	89
6	98
		7	9
8	8

First 1 part of weights is distributed to p _imaximum data transmission unit 6, then 1 part of weights is distributed to p in remaining data transmitting element _imaximum data transmission unit 5.

So obtain each data transmission unit weights n _i(i=1,2,3,4,5,6,7,8), as following table 22.

Table 22 each data transmission unit weights n _i

Data transmission unit	Each data transmission unit integer weights n i
		1	22
2	4
		3	2
4	2
		5	1
6	1
		7	0

8

0

After completing the setting of integer weight of network transmission quantity, carry out verification msg, detailed process is as follows:

Calculate the unit weights diversity factor A of each data transmission unit ₈:

$A_8=\frac{\sqrt{\frac{1}{8}\&times;[{(22-4)}^2+{(4-4)}^2+{(2-4)}^2+{(2-4)}^2+{(1-4)}^2+{(1-4)}^2+{(0-4)}^2+{(0-4)}^2]}}{4}\&ap;1.728$

Calculate similarity S ₈:

$S_8=1-\frac{|A_8-T_8|}{A_8+T_8+1}=1-\frac{|1.728-1.802|}{1.728+1.802+1}\&ap;0.997$

Embodiment is known thus: when data transmission unit quantity is more, method of the present invention still effectively can reduce numerical intervals and ensure higher similarity, therefore avoiding traditional weights method to set up cannot the deficiency of analog network transmission environment truly, improves the efficiency of the network optimization.

Embodiment 3

The present embodiment be the numerical difference of the transmission volume of each data transmission unit apart from very large situation, by the transmission volume of hardware monitoring acquisition four data transfer unit as table 23.

The transmission volume of table 23 embodiment 3 four data transfer unit

Data transmission unit	Transmission volume
		1	25B
2	1KB
		3	13KB
4	1MB

Unified by unit for B, the numerical difference of the transmission volume of each data transmission unit is apart from very large, as shown in table 24 below.

The transmission volume of table 24 embodiment 3 unit four data transfer unit after reunification

Data transmission unit	Transmission volume
		1	25B
2	1024B
		3	13312B
4	1048576B

Computing unit transmission diversity factor T ₄for:

$T_4=\frac{\sqrt{\frac{1}{4}[{(25-265734.25)}^2+{(1024-265734.25)}^2+{(13312-265734.25)}^2+{(1048576-265734.25)}^2]}}{265734.25}\&ap;1.701$

Similar embodiment 1 and embodiment 2, adopt method of the present invention to set the integer weight of network transmission quantity of four data transfer unit in the present embodiment, obtains the numerical intervals of various weight setting method, order of magnitude span and similarity, as following table 25.

The contrast of table 25 conventional method and the inventive method

	Original weights	Sequence number weights	Percentage weights	The present invention
					Data transmission unit 1	25	1	0	1
Data transmission unit 2	1024	2	0	1
					Data transmission unit 3	13312	3	1	2
Data transmission unit 4	1048576	4	99	220
					Numerical intervals	[0,1062973]	[0,10]	[0,100]	[0,224]
Order of magnitude span	6	1	2	2
					Similarity S 4	1	0.602	0.998	0.998

The present invention is in the numerical difference of data transmission unit transmission volume apart from very large situation as can be seen from Table 25, greatly can reduce the numerical intervals of original weights and ensure high similarity effect the same as percentage weights.

Embodiment 4

The present embodiment is the situation that the numerical difference distance of the transmission volume of each data transmission unit is very little, shown in table 26 by the transmission volume of hardware monitoring acquisition four data transfer unit.

The transmission volume of table 26 embodiment 4 four data transfer unit

Data transmission unit	Transmission volume
		1	1800B
2	1810B
		3	1820B
4	1830B

Computing unit transmission diversity factor T ₄for:

$T_4=\frac{\sqrt{\frac{1}{4}[{(1815-1800)}^2+{(1815-1810)}^2+{(1815-1820)}^2+{(1815-1830)}^2]}}{1815}\&ap;0.0062$

Similar embodiment 1 and embodiment 2, adopt method of the present invention to set the integer weight of network transmission quantity of four data transfer unit in the present embodiment, obtains the numerical intervals of various weight setting method, order of magnitude span and similarity, as shown in table 27 below.

The contrast of table 27 conventional method and the inventive method

	Original weights	Sequence number weights	Percentage weights	The present invention
					Data transmission unit 1	1800	1	25	2
Data transmission unit 2	1810	2	25	2
					Data transmission unit 3	1820	3	25	2
Data transmission unit 4	1830	4	25	2
					Numerical intervals	[0,7260]	[1,4]	[0,100]	[0,8]
Order of magnitude span	3	1	2	1
					Similarity S 4	1	0.696	0.988	0.988

Can find out that the numerical difference of the present invention in data transmission unit transmission volume is apart from very little, can use the numerical intervals less than percentage weights and reach high similarity effect the same as percentage weights.

Integrated embodiment 1 ~ 4, outstanding advantages contrast of the present invention is summarized as follows shown in table 28.

Table 28 outstanding advantages contrast table of the present invention

The present invention have employed dexterously and seeks territory algorithm and Q value weighs algorithm surely, the weight values effectively reducing data transmission unit is interval, thus integer weight of network transmission quantity and the such as physical concept such as wrap count, interval number of seconds can be closely linked to use flexibly as optimization of network parameters, without the need to secondary calculating.Such as directly the weights of the present invention's setting can be transmitted within certain period the number of times of data as each data transmission unit, or transmit the interval number of seconds of data as each data transmission unit.Meanwhile, the present invention also assures that weights and the higher similarity of initial data, and up to more than 0.980 (highest similarity is 1), avoiding traditional weights method to set up cannot the deficiency of analog network transmission environment truly.So when multiple stage data transmission unit transmits data simultaneously, the weights that the application of the invention is arranged distribute the limited network bandwidth, more sufficiently and reasonably can utilize network bandwidth resources; When multiple stage data transmission unit transmits data in turn, the weights that the application of the invention is arranged arrange the order of the transmission of data transmission unit in network data, can ensure that network application system can the change of corresponding transmitted data amount efficiently.Therefore the present invention is the establishing method of the integer weights of a set of scientific and effective transmission volume.

Claims (1)

1. an establishing method for integer weight of network transmission quantity, the steps include:

(1) the transmission volume p of m platform data transmission unit is obtained by hardware monitoring _i, wherein i=1, m;

(2) seek territory algorithm: on the basis retaining m platform data transmission unit transmission volume numerical characteristic, find integer region [0, xm], wherein expect that weights x is the weights mathematic expectaion of m platform data transmission unit;

Above-mentioned computational process of seeking territory algorithm is as follows:

1) the upper dividing value of computing unit weights diversity factor for m platform expects that weights are unit weights diversity factor maximum in all weights integrated modes of the data transmission unit of x, the unit weights diversity factor of this integrated mode is maximum, so obtain computing formula as follows:

$max\left(A_m^x\right)=\frac{1}{x}\sqrt{\frac{(m-1){(x-1)}^2+{\&lsqb;(x-1)(m-1)\&rsqb;}^2}{m}}$

Wherein: m is data transmission unit number of units, x is for expecting weights; According to above-mentioned formulae discovery

2) computing unit transmission diversity factor T _m, formula T _mas follows:

$T_m=\frac{\sqrt{\frac{1}{m}\underset{i-1}{\overset{m}{\&Sigma;}}{\left(p_i-\frac{1}{m}\underset{j-1}{\overset{m}{\&Sigma;}}{p}_j\right)}^2}}{\frac{1}{m}\underset{j=1}{\overset{m}{\&Sigma;}}{p}_j}$

Wherein: p _ifor the transmission volume of data transmission unit i; for the averaging network transmission quantity of each data transmission unit; M is data transmission unit number of units;

3) according to the number of units m of data transmission unit with expect weights x, expect that weights x is from little, query steps 1) dividing value in the unit weights diversity factor that calculates find first and be greater than step 2) unit that calculates transmission diversity factor T _munit weights diversity factor on dividing value get corresponding x as expectation weights, obtaining total weight value number is xm, the integer region [0, xm] that the scope that namely have found is suitable;

(3) Q value weighs algorithm surely: utilize Q value in integer region [0, xm], select suitable weights to distribute to m platform data transmission unit, thus calculate the weights of m platform data transmission unit;

The computational process that above-mentioned Q value weighs algorithm is surely as follows:

A) transmission volume of m platform data transmission unit is pressed the ratio of xm calculates;

B) m platform data transmission unit result step a) calculated rounds;

C) deduct step b with total weight value number xm) m platform data transmission unit round after result sum, tentatively remained weights number k, often obtain new weight results in subsequent step and all can refresh k value, if the distribution of k value is over, then obtained the final weights of m platform data transmission unit, algorithm terminates;

If d) have 10 in current weight result just 1 part of weights is distributed to this data transmission unit, k becomes k-1 and returns step c); If having multiple is 0, then 1 part of weights being distributed to weight results is the data transmission unit that in the data transmission unit of 0, transmission volume is maximum, and k becomes k-1 and returns step c); If there is no 0, carry out next step;

E) the Q value of m platform data transmission unit is calculated wherein: n _ifor the weights that each data transmission unit has been distributed, p _ifor the transmission volume of data transmission unit i;

F) 1 part in residue weights is distributed to a maximum data transmission unit of Q value, k becomes k-1 and returns step c).