CN102420747A - Service source shaping method based on date packet achieving interval wave filtration - Google Patents

Abstract

The invention relates to the technical field of service shaping in a mobile communication networking method system, and discloses a service source shaping method based on date packet achieving interval wave filtration. A data packet-based shaping work for achieving the interval wave filtration is within a time domain range, and a smooth object is the time interval at which a data packet is achieved. The performance analysis and the simulation result of an interval filter shaping algorithm (IFSA) indicate that the IFSA has three more outstanding performances: firstly, the method has the characteristic of being low in time delay (the time delay of the data packets are approximately existed); secondly, the method has the characteristic of being good in real-time performance; and the method can be realized without forecasting the service characteristic, thereby being good in realizability.

Description

Service source shaping methods based on the filtering of packet arrival interval

Technical field

The present invention relates to the traffic shaping technical field in the mobile communication network-building method system, be specifically related to a kind of service source shaping methods based on the filtering of packet arrival interval.

Background technology

The mobile network is exactly the network of node motion; Because the mobile uncertainty of introducing network topology structure of node; Thereby the burstiness of professional arrival and the burstiness of data flow have been improved, for mobile network's networking and procotol and system design have brought new technical difficulty.

Traffic shaping is one of main contents of network method system Business Processing, is before business gets into network and professional the entering before certain node, handles arriving professional aggregated flow, to change professional statistical nature.It is low more to arrive professional burstiness, and node state is steady more, thereby network efficiency is high more.Shaping methods divides two types: arrive Business Processing and arrive professional pre-treatment.Arrive Business Processing and be meant that the data flow that arrives business is carried out certain to be handled, and to reduce the burstiness of Business Stream, is referred to as passive shaping methods.Professional arrive pre-treatment and be meant that the application process that business is arrived handles, reducing the professional burstiness that arrives, thereby reduce the burstiness of business data flow.The main shaping of these class methods is to liking long-term business, like IP phone, video conference etc.Their common feature is that the professional duration is longer; Can the service access process be divided into professional application process and service access process; Professional application process is too much consumption of network resources not, and the application process time-delay also can be longer, and to the delay requirement height of transmission course.Can handle professional application process, when not influencing professional transmission course, reduce the burstiness of business polymerization stream.These class methods are called initiatively shaping methods.Because initiatively shaping methods does not influence data transmission procedure, it can combine with other passive shaping methods, accomplishes traffic shaping jointly, to improve shaping effect.

Have the traffic shaping device that designs for atm network at present based on the RC filter.This traffic shaping device has been introduced a kind of and has been leaked the diverse notion of bucket shaping (LB).It is to propose on the LPF theoretical foundation that RC traffic shaping device can carry out the shaping theory to electric current at the RC filter.This method can make output speed along with input rate changes and corresponding the variation through the packet number of filtered method in level and smooth each unit interval.This method mainly contains two weak points.One is in the shaping process, because smooth object is the packet number that arrives in the unit interval, need carry out a certain amount of buffer memory to the packet that arrives, and the length that needs buffer memory is generally greater than 10 packets.Therefore can all introduce a constant time lag Bc/ λ to each packet _In(wherein Bc is the cache size that needs, λ _InBe the average traffic arrival rate).Long so delay time.In addition, this method need know that in realization the peak rate that arrives Business Stream could design suitable reshaper, therefore realizes relatively difficulty.

Summary of the invention

The technical problem that (one) will solve

Technical problem to be solved by this invention is: solve the long and difficult problem of realization of time-delay in traditional passive shaping methods.

(2) technical scheme

For solving the problems of the technologies described above, the invention provides a kind of service source shaping methods based on the filtering of packet arrival interval, may further comprise the steps:

S1, suppose each data packet arrival time of service source { τ _n ^In: n=0,1,2 ... } and describe with random process, earlier before the statistics average arrival time of L packet at interval, as the value of packet average arrival time interval T, obtain T after, think that transmission begins τ _n ^InThe time of advent of representing n packet;

In S2, when beginning transmission,, the timer Timer that record nearest packet time of advent is set is 0, and before being provided with k-1 initial arrival interval for T, T ..., T}, wherein k is the tap coefficient of low pass filter;

If S3 has data arrives; Make data get into buffer earlier; If packet does not get into buffer; Buffer is empty; Then triggering timing device Timer picks up counting, if through minimal buffered duration Tw, next packet does not still arrive; Then the packet in the buffer is sent; And think that once new transmission begins, return execution in step S2, if having new data packets to arrive in the time at Tw; Then the interval time of advent of follow-up data bag is put into successively the intervening sequence time of advent

of i packet and i-1 packet; And carrying out convolution algorithm with low pass filter system function H (ω), the output time intervening sequence

that i packet and i-1 packet are put in the output that obtains successively whenever counts on simultaneously the time of advent of the average at interval of a continuous L packet, just with the value of this average renewal T;

S4, when the numerical value of timer Timer reaches Tw; The output scheduling machine begins successively packet to be exported according to the output gap sequence; Packet of every output; Respectively this packet is removed from the input of correspondence, output time intervening sequence, the machine of output scheduling simultaneously points to packet next to be exported, and newly arrived packet still gets into buffer successively in this process;

If the S5 buffer is empty, then return execution in step S2.

Preferably, said filter is a low pass filter.

Preferably, said low pass filter is a causal system.

Preferably, said convolution algorithm and packet get into parallel the carrying out of process of caching of buffer.

Preferably, among the step S4, the output scheduling machine just triggers the packet output of buffer head of the queue after said buffer has been stored 2 packets at least.

(3) beneficial effect

Compare with RC traffic shaping device, the shaping that the present invention is based on inter-packet gap filtering is operated in the time domain scope, level and smooth to as if time interval of arriving of packet.The performance evaluation of IFSA method and simulation result are shown that IFSA performance aspect three is more outstanding, and at first, this method has low delay characteristic (time-delays of nearly several packets); Secondly, this method has real-time characteristic; The 3rd, when realizing, this method need not predict that realizability is good to traffic performance.

Description of drawings

Fig. 1 is the method flow diagram of the embodiment of the invention;

Fig. 2 is the method implement device structured flowchart of the embodiment of the invention.

Embodiment

Regard to a kind of service source shaping methods proposed by the invention down, specify in conjunction with accompanying drawing and embodiment based on the filtering of packet arrival interval.

As shown in Figure 1, the traffic shaping method based on the filtering of packet arrival interval of the present invention (Interval Filter Shaping Algorithm is called for short IFSA).Compare with RC traffic shaping device, be operated in the time domain scope based on the shaping of inter-packet gap filtering, level and smooth to as if the interval that arrives of packet.The performance evaluation of IFSA method and simulation result are shown that IFSA performance aspect three is more outstanding.At first, this method has low delay characteristic (time-delays of nearly several packets); Secondly, this method has real-time characteristic; The 3rd, when realizing, this method need not predict that realizability is good to traffic performance.Method is described below:

1.1.1 symbol description

(1) average arrival rate of λ (packets/s) packet

(2) T (s/packet) packet average arrival time interval T=1/ λ

(3) B (packets) buffer length

(4) h ₁..., h _kThe tap coefficient of discrete low pass filter (LPF), wherein k is the tap number

(5) frequency domain response of H (ω) LPF

(6) time of advent of i packet

(7) output time of

i packet

(8) time of advent of

i packet and i-1 packet at interval

$X_i^{\mathrm{in}}={\mathrm{\τ}}_i^{\mathrm{in}}-{{\mathrm{\τ}}_{i-1}}^{\mathrm{in}}$

(9) output time of i packet and i-1 packet at interval

$X_i^{\mathrm{out}}={\mathrm{\τ}}_i^{\mathrm{out}}-{{\mathrm{\τ}}_{i-1}}^{\mathrm{out}}$

(10) Timer writes down the timer of nearest packet time of advent

1.1.2 the definition of burstiness

Suppose each data packet arrival time of service source { τ _n ^In: n=0,1,2 ... (wherein, n=0,1,2 ... expression n is the integer more than or equal to 0) can use a random process to describe, then the time interval of each packet arrival constitutes sequence { X _{N '}| X _{N '}=τ _{N '} ^In-τ _{N '-1} ^In: n '=1,2 ... } also be a random process, and the time of advent of service source thus sequence obtain.Therefore, also can a service source be described with the packet intervening sequence time of advent.The burstiness of service source is a parameter that is used for characterizing service source arrival rate intensity of variation or packet interval variation time of advent degree.Different articles all has different definition to burstiness, mainly contains following several kinds:

A. simply describe: β=peak rate/Mean Speed

B. coefficient of variation: $C^2=\frac{\mathrm{Var}\left(X\right)}{E^2\left(X\right)}$

C. diffusion coefficient: $J_n=\frac{\mathrm{Var}(X_{i+1}+...+X_{i+n})}{n{E}^2\left(X\right)}$

Enough information such as frequency, duration and packet correlation that simple description can not provide peak rate to occur.Coefficient of variation C can describe the variation of traffic carrying capacity, it is generally acknowledged that coefficient of variation is the burst information source greater than 1 information source, and its shortcoming is the correlation that can not be used to reflect the adjacent data bag.Diffusion coefficient can obtain the correlation of packet, but this definition makes the analysis more complicated.

Can find out that from above three kinds of sudden descriptions the variance that changes peak rate, packet interval can reduce the burstiness of service source.Consider the accuracy and the complexity of analysis, adopt coefficient of variation to describe burstiness, this also is a kind of describing method of using always.

1.1.3IFSA method is described

IFSA is realized by reshaper and scheduler.Reshaper is made up of input buffer and low pass filter (LPF).The implementation structure figure of IFSA is as shown in Figure 2.ω among the figure ₀It is the cut-off frequency of low pass filter.Tentation data bag bag appearance is same, and the free of data packet loss, and the IFSA method is described below.

Packet at first gets into input buffer after arriving.Timer Timer record each data packet arrival time also obtains the packet intervening sequence time of advent

with

list entries as low pass filter; Corresponding output sequence be exactly packet output time intervening sequence

wherein

be the output time interval of i packet and i+1 packet, output scheduling is just dispatched output according to this at interval.The scheduling output time of i packet is

1.1.4IFSA method implementation

For understanding method realization better, carry out following explanation earlier.In method realizes; Sequence

(i=1; 2; ..., expression i is for more than or equal to 1 integer) be that system function H (ω) convolution by

and low pass filter obtains.This convolution process can be carried out with the process of caching of input is parallel.Because this system is a causal system, promptly has input just output can be arranged, and therefore, when data arrives is arranged, needs the packet of buffer memory some, can exports packet according to output sequence and dispatch.As the minimal buffered number of data packets that triggers output, then the minimum value of N is 2 with basic buffer memory N, and promptly buffer has been stored the packet that could trigger the buffer head of the queue behind 2 packets at least and exported.In realization, substitute basic buffer memory with basic buffer memory duration, then the minimum value of minimal buffered duration Tw=NT is min{Tw}=2T.Wherein T is packet average arrival time interval.T on average obtains (L=10 in the emulation experiment) through the interval stats time of advent to L continuous data bag in implementation procedure.Just can obtain stable output because list entries need pass through certain hour through low pass filter,, when the transmission beginning, suppose to have k-1 input, and be T input time at interval in order to accelerate to stablize.In addition, in realization, buffer is that sky is promptly represented the end of transmission one time, waits for the beginning of transmission next time simultaneously.

The IFSA concrete steps are following:

S1: system initialization process: the average time interval of L packet before adding up earlier, as the value of T.Level and smooth to this L Bao Buzuo, obtain T after, think transmission beginning.

S2: when beginning transmission, Timer=0 is set, and before setting k-1 initial arrival interval be T, T ..., T}.

S3:, be introduced into buffer if data arrives is arranged.If packet does not get into buffer, buffer is empty, and then the triggering timing device picks up counting.If still do not arrive through next packet of Tw time, then the packet in the buffer sent, and think that once new transmission begins, and returns execution in step S2.If have new data packets to arrive in the time at Tw; Just the arrival interval of follow-up data bag is put into time interval sequence

successively and carried out convolution algorithm with filter coefficient; The time of advent of the average at interval that output gap time series whenever counts on a continuous L packet is simultaneously put in the output that obtains successively, just upgraded the value of T with this average.

S4: when Timer=Tw, the output scheduling machine begins successively packet to be exported according to the output gap sequence.Packet of every output is removed this packet respectively from the input of correspondence, output gap sequence, packet next to be exported is pointed in output scheduling simultaneously.Newly arrived packet still gets into buffer successively in this process (process that refers to step S4).

S5:, then return execution in step S2 if buffer is empty.

Can find out that by above embodiment compare with existing RC traffic shaping device, the shaping that the present invention is based on inter-packet gap filtering is operated in the time domain scope, level and smooth to as if time interval of arriving of packet.Performance evaluation and simulation result to IFSA show, IFSA performance aspect three is more outstanding, and at first, this method has low delay characteristic (time-delays of nearly several packets); Secondly, this method has real-time characteristic; The 3rd, when realizing, this method need not predict that realizability is good to traffic performance.

Above execution mode only is used to explain the present invention; And be not limitation of the present invention; The those of ordinary skill in relevant technologies field under the situation that does not break away from the spirit and scope of the present invention, can also be made various variations and modification; Therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.

Claims (5)

1. the service source shaping methods based on the filtering of packet arrival interval is characterized in that, may further comprise the steps:

S1, suppose each data packet arrival time of service source { τ _n ^InN=0,1,2 ... } and describe with random process, earlier before the statistics average arrival time of L packet at interval, as the value of packet average arrival time interval T, obtain T after, think that transmission begins τ _n ^InThe time of advent of representing n packet;

In S2, when beginning transmission,, the timer Timer that record nearest packet time of advent is set is 0, and before being provided with the initial time of advent of k-1 packet be spaced apart T, T ..., T}, wherein k is the tap coefficient of filter;

If S3 has packet to arrive; Make packet get into buffer earlier; If packet does not get into buffer; Buffer is empty; Then triggering timing device Timer picks up counting; If through minimal buffered duration Tw, next packet does not still arrive, and then the packet in the buffer is sent; And think that once new transmission begins; Return execution in step S2,, then the interval time of advent of follow-up data bag is put into successively the intervening sequence time of advent

of i packet and i-1 packet if having new data packets to arrive in the time at Tw; And carry out convolution algorithm with the system function H (ω) of filter; The output time intervening sequence

of i packet and i-1 packet is put in the output that obtains successively, whenever counted on the time of advent of the average at interval of a continuous L packet simultaneously, just with the value of this average renewal T;

S4, when timer Timer numerical value reaches Tw; The output scheduling machine begins successively packet to be exported according to the output time intervening sequence; Packet of every output; Respectively the packet of output is removed from the input of correspondence, output time intervening sequence, the machine of output scheduling simultaneously points to packet next to be exported, and newly arrived packet still gets into buffer successively in this process;

S5, when buffer when being empty, then return execution in step S2.

2. the method for claim 1 is characterized in that, said filter is a low pass filter.

3. the method for claim 1 is characterized in that, said low pass filter is a causal system.

4. the method for claim 1 is characterized in that, said convolution algorithm and packet get into parallel the carrying out of process of caching of buffer.

5. like each described method in the claim 1～4, it is characterized in that among the step S4, the output scheduling machine just triggers the packet output of buffer head of the queue after said buffer has been stored 2 packets at least.

Images