CN103491599A - Tdma protocol prediction control method - Google Patents

Abstract

The invention discloses a TDMA protocol prediction control method which is used for solving the technical problem that an existing TDMA protocol is poor in communication effect. The technical scheme is that feedforward and feedback are combined, the allowed arrival rate of a TDMA communication system is predicted based on a real-time queuing theory model to finish feedforward control, and then the accuracy for the feedforward model to model the complex real-time communication system is further corrected through feedback control. The method effectively avoids the problem of the flow peak value in a TDMA communication system, enables nodes to be joined up quickly and enables system communication to be more stable. Congestion control is dynamically carried out on a TDMA protocol, so that when congestion is detected and happens on communication nodes, time slot resource distribution of node data uses an admission control method to adopt or refuse new service. Thus, the transient flow peak value and network congestion are effectively avoided. The TDMA protocol prediction control method enables the nodes to be joined up quickly, so that the utilization rate of communication resources is higher.

Description

The forecast Control Algorithm of TDMA agreement

Technical field

The present invention relates to a kind of forecast Control Algorithm of TTDMA agreement.

Background technology

In the real time communication environment, because of base station and the travelling carriage scale unstable or unpredictable, travelling carriage go out to network frequent, send the amount of information burst change large, dynamically TDMA receive control more attractive.This agreement can be redistributed time interval resource in real time according to the variation of subscriber signal transmission demand in complex network, has and controls flexibly, the characteristics that the time interval resource utilance is high.

Document 1 " dynamic TDMA resource allocation methods research and implementation. Yan Zhong etc. mobile communication, 2010 (6): 41-44 " a kind of dynamic TDMA agreement based on mini-slot disclosed.The method is in the situation that not interrupt network operation, and each node is applied for time slot according to the actual conditions of information traffic volume and the situation of change of interstitial content, dynamically changes slot length, to adapt to change in topology and to improve the efficiency of transmission of system.

Document 2 " U.S.'s patent of invention of the US6810022B1 of patent publication No. " discloses a kind of TDMA that is applicable to mobile multi-hop Ad-Hoc network and has received control protocol, i.e. unified time slot allocation agreement (USAP).USAP has considered the autonomous behavior of mobile node, allows node to select one or more time slots from unappropriated time slot, by statement and the confirmation to time slot allocation between each adjacent node, assignment information is reached to the double bounce scope.

Document 3 " RFC5352/RFC5356 " discloses a kind of self adaptation time slot distributorship agreement (ASAP).The method, according to the nodes in the double bounce scope, dynamically changes frame length, has reduced untapped timeslot number, has improved channel utilization.Simultaneously, ASAP also provides has the long internodal nothing conflict transmitted in packets of different frame, has effectively utilized channel width.

Yet, above-mentioned dynamic TDMA agreement also has the following disadvantages when solution has the heavy duty tdma communication of clear and definite off period demand: (1) is in the dynamic TDMA agreement of mini-slot, when the message arrival of a plurality of business has exceeded the design load of channel, time interval resource can be deficient, due to instantaneous peak flow, may cause the congested of communication network; (2) in the USAP agreement, the access delay of node is larger, and it adopts the method that peak flow is set to carry out congestion control in addition, and because of congested less generation, this method ensures the tdma communication system with dividing again, and resource utilization is low; (3) the ASAP agreement does not support a plurality of nodes to apply for time slot simultaneously, the network that the business that is only applicable to is constant; And the congestion control policy that it adopts based on direct deduction, have very large randomness and human factor, easily produces error of transmission and loss; And be unsuitable for Accurate Analysis and further improve.

Summary of the invention

In order to overcome the deficiency of existing TDMA protocol communication weak effect, the invention provides a kind of forecast Control Algorithm of TDMA agreement.The method adopts feedforward and feedback to combine, and at first based on real-time queue theory model, the permission arrival rate of prediction tdma communication system, complete feedfoward control; Then by FEEDBACK CONTROL, further correct the inaccuracy of the complicated real-time communication system of feed forward models modeling.The method can effectively be avoided the peak flow problem in the tdma communication system, makes the node access more rapid, and system communication is more sane.

The technical solution adopted for the present invention to solve the technical problems: a kind of forecast Control Algorithm of TDMA agreement is characterized in comprising the following steps:

The feedfoward control unit formed by real-time queue fallout predictor and feedforward controller, the feedback control unit be comprised of congestion detector, feedback controller and timeslot resource management device is realized;

Following hypothesis to business arrival, service time and off period:

(a) business arrives and obeys a renewal process, and arrival interval distributes by average 1/ λ and standard deviation sigma _adetermine;

(b) business has a Random Service time to distribute, by average 1/ μ and standard deviation sigma _sdetermine;

(c) each business has a relative stochastic deadline, obeys G and distributes.Obtain off period mean value by probability theory

a standard knots opinion $\stackrel{\‾}{D}={\∫}_0^{\∞}(1-G\left(x\right))\mathrm{dx};$

(d) the business interval time of advent, separate between service time and off period.

Order

ρ=λ/μ, in real time queueing theory has provided the earliest the preferential approximate deadline missing ratio in the heavy duty situation of off period:

$M_{\mathrm{EDF}}=e^{-\mathrm{\θ}\stackrel{\‾}{D}}---\left(1\right)$

The service rate μ that supposes the tdma communication system is a definite value, and arrival rate λ λ, from the exponential distribution of λ, is derived to formula (1), obtains off period mean value

with miss rate M _eDFmeet following relation:

$M_{\mathrm{EDF}}=e^{-(\frac{2}{\mathrm{\λ}}{\mathrm{\μ}}^2-2\mathrm{\μ})\×\stackrel{\_}{D}}---\left(2\right)$

Accordingly, at known off period mean value

miss rate M with frame of reference _eDFprerequisite under, the arrival rate that the real-time queue fallout predictor allows next control cycle system is predicted:

${\mathrm{\λ}}^{\′}=\frac{2}{2-\frac{\ln M_{\mathrm{EDF}}}{\stackrel{\‾}{D}{\mathrm{\μ}}^2}}---\left(3\right)$

And regulate accordingly arrival rate λ, complete the feedfoward control of system being missed to rate.

The current rate of missing really of congestion detector real-time sampling system system after above-mentioned feedfoward control is missed rate

miss rate M with reference _eDFbetween deviation delta M _eDFto not be 0.

Feedback controller calculates the rate of truly missing

miss rate M with reference _eDFbetween deviation delta M _eDFcalculate again the regulated quantity Δ λ of arrival rate, so that this deviation is as much as possible close to zero.Because within feedfoward control is adjusted to the effective range of FEEDBACK CONTROL by the state of communication system, so consider Δ M during FEEDBACK CONTROL _eDFand the direct mathematical relationship between Δ λ.The arrival rate λ that considers control is the probability distribution of a random process, and main influencing factor is the interval T time of advent _a,

therefore carry out in field of events, reach the change Delta T in the time interval _ato missing rate changes delta M _eDFcontrol.Obtain by the following method interval and miss the relation between the rate time of advent:

1) do not add control algolithm, thereby directly change the rate of missing that the time interval obtains system that arrives;

2) move simulated program, obtain the rate of missing of corresponding system;

3) and then obtain the interval and miss the mathematical relationship between the rate time of advent of system:

$T_a=0.0339M_{\mathrm{EDF}}^{-0.1275}---\left(4\right)$

The time of advent interval change Delta T _awith miss rate changes delta M _eDFbetween relation as follows:

$\mathrm{\Δ}{T}_a\left(n\right)=-0.00432225\×M_{\mathrm{EDF}}{(n-1)}^{-1.1275}\×\mathrm{\Δ}{M}_{\mathrm{EDF}}(n-1)---\left(5\right)$

Use the state of (n-1) time period to be controlled the n time period.

Because the rate of missing is only got on the occasion of with zero, this just makes FEEDBACK CONTROL is a monotonous process, if miss, the arrival rate of system reduces.Because the rate of missing not there will be negative value, the arrival rate of system can not rise all the time, causes the slot efficiency will be more and more lower.To this, then add slot efficiency to system and control, when the slot efficiency of system surpasses a upper limit 0.96, regardless of whether missing, all will adopt the admission control mechanism of front to carry out the utilance of control system.

The timeslot resource management device adopts a saturation block, to the above-mentioned arrival rate calculated

limited:

${\mathrm{\&lambda;}}_q=\left\{\begin{array}{cc}0& \mathrm{\&lambda;}<0\\ \mathrm{\&lambda;}& 0\&le;\mathrm{\&lambda;}\&le;a\\ a& \mathrm{\&lambda;}>a\end{array}\right.---\left(6\right)$

In formula, λ _qfor allowing arrival rate, a is new travelling carriage number of request; Then by λ _qas replying of communication system, send to travelling carriage; Work as λ _q=0 o'clock, travelling carriage did not obtain the data transmission route, works as λ _q, obtain the travelling carriage of data transmission route according to allowing time interval λ at ≠ 0 o'clock _qregulate emission rate.

The invention has the beneficial effects as follows: the method adopts feedforward and feedback to combine, and at first based on real-time queue theory model, the permission arrival rate of prediction tdma communication system, complete feedfoward control; Then by FEEDBACK CONTROL, further correct the inaccuracy of the complicated real-time communication system of feed forward models modeling.The method can effectively be avoided the peak flow problem in the tdma communication system, makes the node access more rapid, and system communication is more sane.

(1) dynamically carry out the congestion control of TDMA agreement with cybernetic method, when communication node detects congested generation, the data slot resource of node is distributed to adopt and receive the method for controlling, receive or refuse new business, thereby effectively having avoided instantaneous peak flow and network congestion.

(2) predict approximate control inputs by queueing theory, when the arrival rate that business detected has a greater change, in advance the permission arrival rate of system is controlled, within always working in its effective range with the feedback controller that guarantees system.This method can make the node access more rapid, and communication resources utilization ratio is higher.

(3) for isomerism, complexity and the unpredictability of communication system, can provide the closed-loop fashion control model.And in the situation that the heavy traffic, there is good behavior definition in the queuing behavior of standard, and the distribution of remaining time off period simultaneously also converges to the form defined.Therefore cybernetics combines with queueing theory and can carry out the accurate Theory analysis to the congested of system, and makes real-time communication system more healthy and stronger.

Below in conjunction with the drawings and specific embodiments, the present invention is elaborated.

The accompanying drawing explanation

Fig. 1 is the tdma communication system feed-forward and feedback control structure based on real-time queueing theory that institute of the present invention method adopts.

Fig. 2 is the tdma communication system group network model that the inventive method adopts.

Fig. 3 is the graph of a relation missed between rate and time of advent interval.

Fig. 4 be not in the same time system add the excitation schematic diagram.

When Fig. 5 is nothing control, system is missed the rate schematic diagram.

Fig. 6 is without control slot utilance schematic diagram.

Fig. 7 misses the rate comparison diagram under different control system.

Fig. 8 is slot efficiency comparison diagram under different control system.

Fig. 9 is that while controlling with/without slot efficiency, system is missed the rate comparison diagram.

Embodiment

Describe the present invention in detail with reference to Fig. 1-9.

A kind of forecast Control Algorithm of TDMA agreement, the feedback control unit that the feedfoward control unit that it mainly is comprised of real-time queue fallout predictor and feedforward controller and congestion detector, feedback controller and timeslot resource management device form forms;

Receive the control process to realize that three kinds of the feedforwards, feedback, feedforward compensation FEEDBACK CONTROL etc. of the forecast Control Algorithm of TDMA agreement receive control modes.

Use OPNET this model of tool modeling and carry out emulation.The network model that emulation is set up consists of a base station and dozens of travelling carriage.Adopting channel data transmission speed during emulation is 30000bps, simulation time is 12h, and the data packet length that node produces is 512bit, the discharge model obeys index distribution that each base station produces, be a constant service time, and application time slot thresholding is set to 100% of queue utilance.High traffic is used in emulation on base station, and packet is produced to the speed average is made as 10pkt/s.

Because heavy duty tdma communication system has clear and definite off period demand, the following hypothesis with reference to real-time queueing theory to business arrival, service time and off period:

(1) business arrives and obeys a renewal process, and arrival interval distributes by average 1/ λ and standard deviation sigma _adetermine;

(2) business has a Random Service time to distribute, by average 1/ μ and standard deviation sigma _sdetermine;

(3) each business has a relative stochastic deadline, obeys G and distributes.Can obtain off period mean value by probability theory

a standard knots opinion $\stackrel{\&OverBar;}{D}={\&Integral;}_0^{\&infin;}(1-G\left(x\right))\mathrm{dx};$

(4) the business interval time of advent, separate between service time and off period.

Order ρ=λ/μ, queueing theory has provided preferential (EDF) approximate deadline missing ratio in the heavy duty situation of off period the earliest in real time:

$M_{\mathrm{EDF}}=e^{-\mathrm{\&theta;}\stackrel{\&OverBar;}{D}}---\left(1\right)$

The service rate μ that supposes the tdma communication system is a definite value, and arrival rate λ obeys the exponential distribution of λ, and formula (1) is derived, and obtains off period mean value

with miss rate M _eDFmeet following relation:

$M_{\mathrm{EDF}}=e^{-(\frac{2}{\mathrm{\&lambda;}}{\mathrm{\&mu;}}^2-2\mathrm{\&mu;})\&times;\stackrel{\&OverBar;}{D}}---\left(2\right)$

Accordingly, at known off period mean value

miss rate M with frame of reference _eDFprerequisite under, the arrival rate that the real-time queue fallout predictor allows next control cycle system is predicted:

${\mathrm{\&lambda;}}^{\&prime;}=\frac{2}{2-\frac{\ln M_{\mathrm{EDF}}}{\stackrel{\&OverBar;}{D}{\mathrm{\&mu;}}^2}}---\left(3\right)$

And regulate accordingly arrival rate λ, complete the feedfoward control of system being missed to rate.

The current rate of missing really of congestion detector real-time sampling system

due to the inaccuracy of the complicated real-time communication system of queue theory model modeling, the system after above-mentioned feedfoward control is missed rate

miss rate M with reference _eDFbetween deviation delta M _eDFto not be 0.

Feedback controller calculates the rate of truly missing

miss rate M with reference _eDFbetween deviation delta M _eDF, then calculate the regulated quantity Δ λ of arrival rate, so that this deviation is as much as possible close to zero.Because within feedfoward control is adjusted to the effective range of FEEDBACK CONTROL by the state of communication system, so consider Δ M during FEEDBACK CONTROL _eDFand the direct mathematical relationship between Δ λ.The arrival rate λ that considers control is the probability distribution of a random process, and main influencing factor is the interval T time of advent _a, therefore carry out in field of events, reach the change Delta T in the time interval _ato missing rate changes delta M _eDFcontrol.Obtain by the following method interval and miss the relation between the rate time of advent:

1) do not add control algolithm, thereby directly change the rate of missing that the time interval obtains system that arrives;

2) move simulated program, obtain the rate of missing of corresponding system;

3) and then can obtain the interval and miss the mathematical relationship between the rate time of advent of system:

$T_a=0.0339M_{\mathrm{EDF}}^{-0.1275}---\left(4\right)$

So, the time of advent interval change Delta T _awith miss rate changes delta M _eDFbetween relation as follows:

ΔT _a(n)＝-0.00432225×M _EDF(n-1) ^-1.l275×ΔM _EDF(n-1) (5)

Use the state of (n-1) time period to be controlled the n time period.

Because the rate of missing is only got on the occasion of with zero, this just makes FEEDBACK CONTROL is a monotonous process, if miss, the arrival rate of system reduces.Because the rate of missing not there will be negative value, the arrival rate of system can not rise all the time, causes the slot efficiency will be more and more lower.To this, then add slot efficiency to system and control, when the slot efficiency of system surpasses a upper limit 0.96, regardless of whether missing, all will adopt the admission control mechanism of front to carry out the utilance of control system.

The timeslot resource management device adopts a saturation block, to the above-mentioned arrival rate calculated

limited:

${\mathrm{\&lambda;}}_q=\left\{\begin{array}{cc}0& \mathrm{\&lambda;}<0\\ \mathrm{\&lambda;}& 0\&le;\mathrm{\&lambda;}\&le;a\\ a& \mathrm{\&lambda;}>a\end{array}\right.---\left(6\right)$

In formula, λ _qfor allowing arrival rate, a is new travelling carriage number of request; Then by λ _qas replying of communication system, send to travelling carriage; Work as λ _q=0 o'clock, travelling carriage did not obtain the data transmission route, works as λ _q, obtain the travelling carriage of data transmission route according to allowing time interval λ at ≠ 0 o'clock _qregulate emission rate.

As can be seen from Figure 5, when adding again the step of 30% to an approximate saturated system, in uncontrolled situation, system will soon be saturated, and due to the message had little time in the treatment system queue, the rate of missing of system will be very high, reach 100%, cause system congestion.

As can be seen from Figure 6, reaction is on slot efficiency, and system still reaches saturated very soon, causes system congestion.

As can be seen from Figure 7, when only adding feedfoward control, corresponding to the feedforward part, the service rate μ that supposes the tdma communication system is a definite value, arrival rate λ obeys the exponential distribution of λ, preferential (EDF) off period mean value in the heavy duty situation of the off period the earliest that feedfoward control provides based on real-time queueing theory

with miss rate M _eDFbetween approximation relation:

$M_{\mathrm{EDF}}=e^{-(\frac{2}{\mathrm{\&lambda;}}{\mathrm{\&mu;}}^2-2\mathrm{\&mu;})\&times;\stackrel{\_}{D}}$

The known reference system is missed rate M _eDF, and obtain off period mean value by the Real-Time Monitoring communication system

, the arrival rate that the real-time queue fallout predictor allows next control cycle system is predicted:

${\mathrm{\&lambda;}}^{\&prime;}=\frac{2}{2-\frac{\ln M_{\mathrm{EDF}}}{\stackrel{\&OverBar;}{D}{\mathrm{\&mu;}}^2}}$

And regulate accordingly arrival rate λ.With without controlling, compare, be very significantly improved through the performance of feedforward control system, but, due to the inexactness of forecast model, cause simple feedfoward control not ideal enough.

Feedback controller is a proportional integral (PI) controller, under periodic control interval h, utilizes the rate of missing of TELE COMMUNCATION SYSTEM actual measurement

with reference value M _eDFbetween difference DELTA M _eDF, calculate kh new arrival rate λ (kh) constantly according to following discrete time control law:

$\mathrm{\&lambda;}\left(\mathrm{kh}\right)=\mathrm{K\&Delta;}{M}_{\mathrm{EDF}}\left(\mathrm{kh}\right)+\underset{i=0}{\overset{k-1}{\mathrm{\&Sigma;}}}\frac{K}{T_i}\mathrm{\&Delta;}{M}_{\mathrm{EDF}}\left(\mathrm{ih}\right)$

By Digital PID Controller Ziegler-Nichols method, utilize Matlab can obtain the parameter of PI controller: K=16, T _i=1.8.So because simple feedback controller be by the accumulation of error in the past to communication system behavior in the future controlled and certainly had certain delay, and it is also undesirable to control effect when large step is arranged, maximum is missed rate and is reached 50%.

Feedforward and feedback are combined, utilize feedfoward control to carry out the state of prognoses system, timely adjustment System state, within adjusting to the state of communication system the effective range of FEEDBACK CONTROL, the mathematical relationship of directly utilizing formula (5) to set up, by the change Delta T at the interval time of advent _ato missing rate changes delta M _eDFcarry out FEEDBACK CONTROL, make when step is arranged the time delay of FEEDBACK CONTROL less, simultaneously, FEEDBACK CONTROL has compensated again the inexactness of feedfoward control.

As can be seen from Figure 8, the slot efficiency of FEEDBACK CONTROL is lower, and its reason is that to miss rate be a special variable, and it is only got on the occasion of with zero.This just makes FEEDBACK CONTROL is the process of a dullness, if miss, the arrival rate of system reduces, and because the rate of missing not there will be negative value, the arrival rate of system can not rise all the time, causes slot efficiency more and more lower.So it is infeasible only by FEEDBACK CONTROL, carrying out the rate of missing of control system in practice.

As can be seen from Figure 9, due to the defect of FEEDBACK CONTROL self, only have to miss and just can be controlled afterwards, the rate of missing like this has some little fluctuations certainly, in Fig. 7, has some burrs to exist.Below adding a slot efficiency control to system again will make systematic function better.Whether specific practice is: when the slot efficiency of system surpasses one upper (in experiment, being decided to be 0.96) in limited time, no matter miss, all will adopt the admission control mechanism of front to carry out the utilance of control system.The fluctuation that can find out the rate of missing of the system that has added slot efficiency control obviously reduces.

Claims (1)

1. the forecast Control Algorithm of a TDMA agreement is characterized in that comprising the following steps:

The feedfoward control unit formed by real-time queue fallout predictor and feedforward controller, the feedback control unit be comprised of congestion detector, feedback controller and timeslot resource management device is realized;

Following hypothesis to business arrival, service time and off period:

(a) business arrives and obeys a renewal process, and arrival interval distributes by average 1/ λ and standard deviation sigma _adetermine;

(b) business has a Random Service time to distribute, by average 1/ μ and standard deviation sigma _sdetermine;

(c) each business has a relative stochastic deadline, obeys G and distributes; Obtain off period mean value by probability theory

a standard knots opinion $\stackrel{\&OverBar;}{D}={\&Integral;}_0^{\&infin;}(1-G\left(x\right))\mathrm{dx};$

(d) the business interval time of advent, separate between service time and off period;

Order

ρ=λ/μ, in real time queueing theory has provided the earliest the preferential approximate deadline missing ratio in the heavy duty situation of off period:

$M_{\mathrm{EDF}}=e^{-\mathrm{\&theta;}\stackrel{\&OverBar;}{D}}---\left(1\right)$

The service rate μ that supposes the tdma communication system is a definite value, and arrival rate λ obeys the exponential distribution of λ, and formula (1) is derived, and obtains off period mean value

with miss rate M _eDFmeet following relation:

$M_{\mathrm{EDF}}=e^{-(\frac{2}{\mathrm{\&lambda;}}{\mathrm{\&mu;}}^2-2\mathrm{\&mu;})\&times;\stackrel{\&OverBar;}{D}}---\left(2\right)$

Accordingly, at known off period mean value

miss rate M with frame of reference _eDFprerequisite under, the arrival rate that the real-time queue fallout predictor allows next control cycle system is predicted:

${\mathrm{\&lambda;}}^{\&prime;}=\frac{2}{2-\frac{\ln M_{\mathrm{EDF}}}{D{\mathrm{\&mu;}}^2}}---\left(3\right)$

And regulate accordingly arrival rate λ, complete the feedfoward control of system being missed to rate;

The current rate of missing really of congestion detector real-time sampling system system after above-mentioned feedfoward control is missed rate miss rate M with reference _eDFbetween deviation delta M _eDFto not be 0;

Feedback controller calculates the rate of truly missing

miss rate M with reference _eDFbetween deviation delta M _eDF, then calculate the regulated quantity Δ λ of arrival rate, so that this deviation is as much as possible close to zero; Because within feedfoward control is adjusted to the effective range of FEEDBACK CONTROL by the state of communication system, so consider Δ M during FEEDBACK CONTROL _eDfand the direct mathematical relationship between Δ λ; The arrival rate λ that considers control is the probability distribution of a random process, and main influencing factor is the interval T time of advent _a,

therefore carry out in field of events, reach the change Delta T in the time interval _ato missing rate changes delta M _eDFcontrol; Obtain by the following method interval and miss the relation between the rate time of advent:

1) do not add control algolithm, thereby directly change the rate of missing that the time interval obtains system that arrives;

2) move simulated program, obtain the rate of missing of corresponding system;

3) and then obtain the interval and miss the mathematical relationship between the rate time of advent of system:

$T_a=0.0339M_{\mathrm{EDF}}^{-0.1275}---\left(4\right)$

The time of advent interval change Delta T _awith miss rate changes delta M _eDFbetween relation as follows:

ΔT _a(n)＝-0.00432225×M _EDF(n-1) ^-1.1275×ΔM _EDF(n-1) (5)

Use the state of (n-1) time period to be controlled the n time period;

Because the rate of missing is only got on the occasion of with zero, this just makes FEEDBACK CONTROL is a monotonous process, if miss, the arrival rate of system reduces; Because the rate of missing not there will be negative value, the arrival rate of system can not rise all the time, causes the slot efficiency will be more and more lower; To this, then add slot efficiency to system and control, when the slot efficiency of system surpasses a upper limit 096, regardless of whether missing, all will adopt the admission control mechanism of front to carry out the utilance of control system;

The timeslot resource management device adopts a saturation block, to the above-mentioned arrival rate calculated

limited:

${\mathrm{\&lambda;}}_q=\left\{\begin{array}{cc}0& \mathrm{\&lambda;}<0\\ \mathrm{\&lambda;}& 0\&le;\mathrm{\&lambda;}\&le;a\\ a& \mathrm{\&lambda;}>a\end{array}\right.---\left(6\right)$

In formula, λ _qfor allowing arrival rate, a is new travelling carriage number of request; Then by λ _qas replying of communication system, send to travelling carriage; Work as λ _q=0 o'clock, travelling carriage did not obtain the data transmission route, works as λ _q, obtain the travelling carriage of data transmission route according to allowing time interval λ at ≠ 0 o'clock _qregulate emission rate.

Images