WO2023109007A1

WO2023109007A1 - Time domain resource configuration method and apparatus, electronic device, and storage medium

Info

Publication number: WO2023109007A1
Application number: PCT/CN2022/093661
Authority: WO
Inventors: 丰雷; 李卫; 王建全; 孙雷; 周雨; 刘珊; 黄蓉; 谢坤宜; 周彦伯
Original assignee: 北京邮电大学; 北京科技大学
Priority date: 2021-12-17
Filing date: 2022-05-18
Publication date: 2023-06-22
Also published as: CN114501644A

Abstract

The present application provides a time domain resource configuration method and apparatus, an electronic device, and a storage medium. The method comprises: obtaining an error covariance of clock synchronization precision; constructing a time domain resource configuration optimization problem on the basis of the error covariance of the clock synchronization precision, a time-preserving band length, and a data packet arrival rate, wherein the time domain resource configuration optimization problem is a problem of maximizing network throughput under the condition of ensuring the clock synchronization precision; solving the time domain resource configuration optimization problem to obtain the data packet arrival rate; and determining the network throughput on the basis of the data packet arrival rate and the time-preserving band length. According to the time domain resource configuration method and apparatus, the electronic device, and the storage medium provided in the present application, the maximized network throughput is calculated on the basis of the time-preserving band length and the data packet arrival rate obtained by solving the time domain resource configuration optimization problem, and the maximized network throughput reflects the configuration of time domain resources, so that optimization of the time domain resource configuration is achieved.

Description

Time-domain resource allocation method, device, electronic device and storage medium

cross reference

This application refers to the Chinese patent application No. 2021115554221 filed on December 17, 2021 with the patent title "Time Domain Resource Configuration Method, Device, Electronic Equipment, and Storage Medium", which is fully incorporated by reference into this application.

technical field

The present application relates to the technical field of wireless communication, and in particular to a time domain resource configuration method, device, electronic equipment and storage medium.

Background technique

The 5th generation mobile network (5G) technology can provide low-latency and ultra-reliable deterministic functions for industrial control systems and data traffic, and its wireless technology can provide flexible system connections for entire industrial manufacturing facilities and factories. sex. Time Sensitive Network (TSN) is a series of IEEE 802 Ethernet standards formulated by the IEEE 802.1TSN task group, which can provide the network with the ability of low delay, low jitter and extremely low data loss rate, so that Ethernet can It is suitable for time-sensitive application scenarios with strict reliability and delay requirements. As 5G for wireless communication, connecting industrial equipment such as sensors and actuators to the TSN network in a wireless manner can give full play to the flexibility of 5G and the extremely low latency of TSN, effectively reducing cable laying and the wide application of mobile devices. .

The coexistence of 5G and TSN networks puts forward higher requirements for 5G deterministic communication capabilities, which not only require 5G networks to provide Ultra Reliable Low Latency Communications (URLLC) capabilities, but also require synchronous clocks provided through 5G air interfaces Complete the timing. It is an urgent problem to be solved at present to evaluate the determinism of delay in an uncertain wireless link environment, so as to further ensure the accuracy of clock synchronization as a support for time domain resource allocation.

Contents of the invention

Aiming at the problems existing in the prior art, the present invention provides a time domain resource configuration method, device, electronic equipment and storage medium.

The present invention provides a time domain resource configuration method, including:

Obtain the error covariance of clock synchronization accuracy;

Based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length and the arrival rate of data packets, the time-domain resource allocation optimization problem is constructed; wherein, the time-domain resource allocation optimization problem is to make the network throughput under the condition of ensuring the clock synchronization accuracy The problem of maximizing the quantity;

Solving the time-domain resource allocation optimization problem to obtain the data packet arrival rate;

The network throughput is determined based on the data packet arrival rate and the length of the timekeeping band.

According to a time-domain resource configuration method provided by the present invention, the error covariance of obtaining clock synchronization accuracy includes:

Construct a timestamp observation model based on timestamp information and clock state variables;

Evaluating network delay certainty based on the time stamp observation model;

The error covariance of the clock synchronization accuracy is determined based on the evaluated timestamp observation model.

According to a time-domain resource configuration method provided by the present invention, the construction of a time stamp observation model based on time stamp information and clock state variables includes:

Build the time stamp observation model based on formula (1);

z _i,k =γ _k [T _i,1 +T _i,2 -2t]=γ _k [Cx _i (k)+v _i (k)] (1)

Among them, z _i,k is the time stamp observation model, γ _k is a binary Bernoulli random variable, i is the number of base station nodes, T _i,1 is the time synchronization sent by base station node gNB _i to timing terminal node UE _j The moment of the variable, T _i,2 is the moment when the base station node gNB _i receives the reply time synchronization variable from the terminal node UE _j , j is the number of terminal nodes, t is the ideal reference time, C is the observation matrix, C=[0 2] , x _i (k) is the clock state variable of the base station node gNB _i , k is the kth period of the clock synchronization process, v _i (k) is the observation noise, v _i (k)=d _i +Y _i,k , v _i (k) is a Gaussian random variable that obeys the N(0,R) distribution, d _i is the fixed time delay generated during the synchronization process, and Y _{i, k} is the random time delay generated due to the randomness of the wireless channel.

According to a time-domain resource configuration method provided by the present invention, the evaluation of network delay certainty based on the time stamp observation model includes:

Solving the confidence degree of the wireless network delay certainty in the formula (2);

p _d ＝p{t _l <x<αt _u }p{0<w<(1-α)t _u } (2)

Among them, p _d is the confidence degree of the wireless network delay determinism, α is the delay scaling factor, 0<α<t _l /t _u , t _l is the lower bound of the sum of transmission delay and queuing delay, t _u is the upper bound of the sum of transmission delay and queuing delay, x is the random variable of transmission delay, w is the random variable of queuing delay of nodes in the wireless network; p{t _l <x<αt _u } is the Probability value between t _l and αt _u ;

z(t)=(2 ^L/(Wt) -1)/c, L is the length of the transmitted data packet, W is the bandwidth of the channel, and c is the constant coefficient of the receiving SNR; p{0<w<(1 -α)t _u } is the probability value of w falling between 0 and (1-α)t _u ;

θ satisfies the constraint condition E[e ^θA(1) ]E[e ^-θS(1) ]≤1, A(t) is the amount of data arrival in (0,t) period, S(t) is (0,t ) data service volume in the time period; A(1) is the data arrival volume per unit time, S(1) is the data service volume per unit time, and S[(1-α)t _u ] is (0,(1 -α)t _u ) data service volume in the time period;

The distribution of γ _k is determined based on the confidence of the wireless network delay determinism.

According to a time-domain resource configuration method provided by the present invention, the determination of the error covariance of the clock synchronization accuracy based on the evaluated time stamp observation model includes:

Determine the error covariance of the clock synchronization accuracy based on formula (3);

P _k ＝E[e _k e _k ^T ] (3)

Wherein, P _k is ^{the error covariance of the clock synchronization accuracy, ek is the estimation error of the clock state variable, ek T} _is _the transposition matrix of _ek ,

x _k is the clock state variable,

is the estimation of x _k under the condition that the observed value is known,

According to a time-domain resource allocation method provided by the present invention, the time-domain resource allocation optimization problem based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length, and the data packet arrival rate is expressed as:

P1:max{U(B,λ)};

C2: Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C3: 0<λ<1;

Wherein, P1 is the objective function of the time-domain resource allocation optimization problem, C1 to C3 are the constraints of the objective function P1, U(B,λ)=R _b , B is the length of the time-keeping band, and λ is the Packet arrival rate, R _b is the network throughput, P _k is the error covariance of the clock synchronization accuracy, E[P _k ] is the expected value of P _k , E[P _k ]=AXA ^T +Q-λAXC ^T ( CXC ^T +R) ^-1 CXA ^T , X=E[P _k-1 ], λ=E[γ _k ]=p _d , ΔE _k is the fixed error in the synchronization process, Tr(E[P _k ]) is The trace of E[P _k ], T _S is the time slot length of the data packet, A is the coefficient matrix, A ^T is the transpose matrix of the coefficient matrix, Q is the covariance matrix of the process noise, R is the covariance of the observation noise, P _k-1 is the error covariance matrix at time k-1.

According to a time-domain resource configuration method provided by the present invention, the solving of the time-domain resource configuration optimization problem to obtain the data packet arrival rate includes:

Convert the time-domain resource allocation optimization problem into a convex optimization problem:

P2:min{Tr(P _k )E[a _k |P _k ]-VU}

stC4:Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C5:0<λ<1;

Among them, P2 is the objective function of the convex optimization problem, C4 and C5 are the constraints of the objective function P2, Tr(P _k ) is the trace of P _k , a _k is the random arrival process of the virtual queue, E[a _k | P _k ] is the expected value of a _k when P _k is known, V is the control coefficient,

P _k+1 ＝max{P _k -b _k ,0}+a _k , b _k is the departure process of the virtual queue, b _k is a constant, and b _k ≥ E[a _k ];

Solving the convex optimization problem to obtain the data packet arrival rate.

According to a time-domain resource configuration method provided by the present invention, the determining the network throughput based on the arrival rate of the data packet and the length of the time-guaranteed band includes:

Determine the network throughput based on formula (4);

Among them, W _e is the equivalent bandwidth of the time-guaranteed band length,

is the inverse function of Q(x),

1-λ is the probability of packet loss, and t is the integral variable.

The present invention also provides a device for configuring time-domain resources, including:

An acquisition unit, configured to acquire the error covariance of the clock synchronization accuracy;

A construction unit for constructing a time-domain resource configuration optimization problem based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length, and the arrival rate of data packets; wherein, the time-domain resource configuration optimization problem is to ensure clock synchronization accuracy The problem of maximizing network throughput under the condition;

A solving unit, configured to solve the time-domain resource allocation optimization problem to obtain the data packet arrival rate;

A determining unit, configured to determine the network throughput based on the arrival rate of the data packet and the length of the time guarantee band.

The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and operable on the processor. When the processor executes the program, it realizes any of the above-mentioned time-domain resources. Steps to configure the method.

The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the time-domain resource allocation methods described above are implemented.

The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of any one of the time-domain resource allocation methods described above are implemented.

A time-domain resource allocation method, device, electronic equipment, and storage medium provided by the present invention construct a time-domain resource allocation optimization problem based on the error covariance of clock synchronization accuracy, the length of the time-guarantee band, and the arrival rate of data packets, and based on the guarantee The maximum network throughput can be obtained by calculating the time band length and the data packet arrival rate obtained by solving the time domain resource allocation optimization problem. Since the maximized network throughput reflects the allocation of time domain resources, the optimization of time domain resource allocation is realized.

Description of drawings

In order to more clearly illustrate the technical solutions in this application or the prior art, the accompanying drawings that need to be used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings in the following description are the present For some embodiments of the application, those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is one of the schematic flow charts of the time domain resource allocation method provided by the present application;

FIG. 2 is the second schematic flow diagram of the time domain resource allocation method provided by the present application;

Fig. 3 is a schematic diagram of the time stamp observation model of the two-way information exchange mechanism provided by the present application;

FIG. 4 is a schematic structural diagram of a time-domain resource configuration device provided by the present application;

FIG. 5 is a schematic diagram of the physical structure of the electronic device provided by the present application.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention , but not all examples. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

The time domain resource configuration method of the present invention will be described below with reference to FIG. 1 to FIG. 3 .

Fig. 1 is one of the flow diagrams of the time domain resource configuration method provided by the present invention. As shown in Fig. 1, the time domain resource configuration method includes:

Step 101. Obtain error covariance of clock synchronization accuracy.

Step 102, constructing a resource allocation optimization problem in the time domain based on the error covariance of the clock synchronization accuracy, the length of the time-guaranteed band, and the arrival rate of data packets.

Wherein, the time-domain resource configuration optimization problem is a problem of maximizing network throughput under the condition of ensuring clock synchronization accuracy.

Step 103, solving the time-domain resource allocation optimization problem to obtain the data packet arrival rate.

Step 104. Determine the network throughput based on the data packet arrival rate and the time-keeping band length.

A time-domain resource configuration method provided by the present invention constructs a time-domain resource allocation optimization problem based on the error covariance of clock synchronization accuracy, the length of the time-guaranteed band and the arrival rate of data packets, and solves the time-domain resource based on the length of the time-guaranteed band and the The packet arrival rate calculation obtained from the configuration optimization problem maximizes the network throughput. Since the maximized network throughput reflects the allocation of time domain resources, the optimization of time domain resource allocation is realized.

Optionally, FIG. 2 is the second schematic flow diagram of the time domain resource configuration method provided by the present invention. As shown in FIG. 2, step 101 in FIG. 1 can be specifically implemented through the following steps:

Step 1011, construct a time stamp observation model based on the time stamp information and clock state variables.

Optionally, construct the time stamp observation model based on formula (1):

z _i,k =γ _k [T _i,1 +T _i,2 -2t]=γ _k [Cx _i (k)+v _i (k)] (1)

Among them, z _i,k is the time stamp observation model, γ _k is a binary Bernoulli random variable, which can also be called the time observation variable value, i is the number of base station nodes, T _i,1 is the base station node gNB _i The time when the time synchronization variable is sent to the time service terminal node UE _j , T _i,2 is the time when the base station node gNB _i receives the time synchronization variable from the terminal node UE _j , j is the number of terminal nodes, t is the ideal reference time, and C is Observation matrix, C=[0 2], x _i (k) is the clock state variable of base station node gNB _i , k is the kth cycle of the clock synchronization process, v _i (k) is the observation noise, v _i (k) = d _i +Y _i,k , v _i (k) is a Gaussian random variable that obeys the N(0,R) distribution, d _i is the fixed delay generated during the synchronization process, and Y _i,k is due to the randomness of the wireless channel Generated random delays.

Exemplarily, the present invention is applied to a wireless communication environment composed of a 5G base station and a time service terminal, and a clock synchronization system is formed between the 5G base station and the time service terminal through two-way information exchange measurement, then the recurrence equation of the clock state variable of the base station node gNB _i can be Expressed by the following formula (5):

x _i (k) = _Axi (k－1) + w _i (k) + b (5)

in,

β _i (k) is the instantaneous clock skew of gNB _i , that is, the frequency offset of the clock synchronization signal;

is the cumulative clock offset of gNB _i , that is, the phase offset; x _i (k-1) is the clock state variable of base station node gNB _i at time k-1; A is the coefficient matrix,

τ ₀ is the sampling period; w _i (k) is the process noise, w _i (k)=[μ _i (k)τ ₀ μ _i (k)] ^T ;

p _i is the phase noise parameter; B _i (k) is the standard Wiener process; B _i ′(k) is the difference of B _i (k) with respect to k; B _i ′(k-1) is the difference of B _i (k-1 ) about the difference of k-1; B _i (k-1) is the standard Wiener process of the k-1th period; μ _i (k) is a Gaussian process with a mean of 0 and a variance of 2p _i ; b is a constant matrix , b=[0-τ ₀ ] ^T ; w _i (k) satisfies the following formula (6) and formula (7):

E[w _i (k)]=0 (6)

E[w _i (k)w _i ^T (k)]=Q (7)

Among them, E[w _i (k)] is the expected value of w _i (k), and E[w _i (k)w _i ^T (k)] is the expected value of w _i (k)w _i ^T (k).

Based on the above definition of clock state variables, a time stamp observation model based on a two-way information exchange mechanism is established. Fig. 3 is a schematic diagram of the time stamp observation model of the two-way information exchange mechanism provided by the present invention. As shown in Fig. 3, it is assumed that the time stamp set of the base station node gNB _i in a two-way clock synchronization process is {T _{i, 1} , T _{i ,2} }, where T _i,1 is the moment when the base station node gNB _i sends the time synchronization variable to the timing terminal node UE _j , T _i,1 is expressed by the following formula (8), and T _i,2 is the time when the base station node gNB _i receives The moment when the terminal node UE _j replies to the time synchronization variable, T _i,2 is expressed by the following formula (9):

in,

is the accumulated clock offset of gNB _i , d _i is the fixed time delay generated in the synchronization process, Y _i,k is the random time delay due to the randomness of the wireless channel, t=kτ ₀ represents the ideal reference time, due to the two-way information The time taken by the exchange is short compared to the time required by the entire clock synchronization algorithm, so the ideal reference time t can be considered to remain constant in a round of two-way information exchange; studies have shown that such delays can be modeled as Gaussian random process of the distribution.

It should be noted that the request packet in Figure 3 is a data packet sent by the base station node gNB _i to the timing terminal node UE _j , and the reply packet is a data packet replied by the terminal node UE _j to the base station node gNB _i .

The initial observation model of the 5G base station node gNB _i is obtained by adding formula (8) and formula (9) and transforming it. The initial observation model is expressed by the following formula (10):

Among them, y _i,k is the initial observation model.

combined with the above

The initial observation model can be expressed by the following formula (11):

y _i,k =Cx _i (k)+v _i (k) (11)

Wherein, observation matrix C=[0 2], observation noise v _i (k)=d _i +Y _{i, k} is a Gaussian random variable obeying N(0,R) distribution.

In the 5G network, due to the influence of the random variables of wireless channel fading, the transmission of the link is no longer reliable, that is, during the two-way information exchange process of clock synchronization, gNB _i sends timestamp information to UE _j and UE _j replies to gNB The timestamp information of _i cannot be guaranteed to arrive. Therefore, it is considered to modify the initial observation model of formula (11) to make it suitable for modeling the time synchronization mechanism in the wireless industrial network scenario. The corrected initial observation model is the time stamp observation model expressed by the above formula (1).

Step 1012: Evaluate the network delay certainty based on the time stamp observation model.

Optionally, the confidence degree of wireless network delay certainty in formula (2) is solved; and the distribution of γ _k is determined based on the confidence degree of wireless network delay certainty.

p _d ＝p{t _l <x<αt _u }p{0<w<(1-α)t _u } (2)

θ satisfies the constraint condition E[e ^θA(1) ]E[e ^-θS(1) ]≤1, A(t) is the amount of data arrival in (0,t) period, S(t) is (0,t ) data service volume in the time period,

for

The expected value of , E[e ^θA(1) ] is the expected value of e ^θA(1) , E[e ^-θS(1) ] is the expected value of e ^-θS(1) .

As an example, the descriptive index of 5G wireless network determinism is discussed based on binary Bernoulli random variable γ _k . For the binary Bernoulli random variable γ _k , the probability distribution of its value is reflected by the probability that the time delay in the wireless network falls within the specified interval, that is, γ _k ~ b(1,p _d ), p _d adopts the following formula (12) means:

p _d ＝p{t _l <x+w<t _u },t _l <t _u (12)

Among them, b is the Bernoulli distribution, x represents the random variable of the transmission delay, and its randomness is mainly caused by the fading variable of the wireless channel; the random variable of the queuing delay of nodes in the wireless network is determined by the arrival process of the data packet and The joint random variable jointly determined by the random process of channel fading; t _l is the lower bound of the time delay, t _u is the upper bound of the time delay; the value of p _d is equal to the value of x+w and falls within the interval (t _l , t _u ) Probability value between . Generally speaking, the transmission delay and queuing delay of data packets are the main components of the node delay in the communication system, so the value of p _d can more accurately describe the confidence of the delay deterministic estimation in the wireless deterministic network. Since w is a joint random variable, it is difficult to solve formula (12) directly, so the calculation formula of p _d is updated to the above formula (2). Obviously, the time delay boundary condition of formula (2) is a sufficient deficiency of formula (12). The necessary condition, that is, when the time-delay random variable satisfies the probability condition of formula (2), it must satisfy the time-delay boundary of formula (12), and vice versa. Therefore, formula (2) is used as the value of p _d to reflect the degree of certainty of the time delay estimation, and then describe the distribution of the binary Bernoulli random variable γ _k .

The solution of p _d in formula (2) can be divided into two parts x and w. Regarding the part x, since x is a function of the random variable of channel fading, for a Rayleigh random channel with parameter σ, x in formula (2) The interval probability of can be obtained by the Shannon channel capacity formula and the cumulative distribution function of the channel random variable, expressed by formula (13):

Wherein, z (t)=(2 ^L/(Wt) -1)/c, L is the length of sending data packet, W is the bandwidth of channel, and c is the constant coefficient of receiving signal-to-noise ratio;

Regarding the part w, the random network calculus theory based on the moment generating function (Moment Generating Function, MGF) is used to solve the interval probability of w in the formula (2). In the service process, the upper boundary of w satisfies the following formula (14):

Among them, θ satisfies the constraint condition E[e ^θA(1) ]E[e ^-θS(1) ]≤1, A(t) is the data arrival amount in (0,t) period, S(t) is (0 , t) the data service volume in the time period; A(1) is the data arrival volume per unit time, S(1) is the data service volume per unit time, and S[(1-α)t _u ] is (0, (1-α)t _u ) data service volume in the time period; when the value of θ makes the formula E[e ^θA(1) ]E[e ^-θS(1) ]≤1 approximately 1, then the formula (14 ) is approximately established, from which the interval probability of w in formula (2) can be obtained, and then the value of p _d can be obtained to complete the determination of the binary Bernoulli random variable γ _k distribution.

Step 1013: Determine the error covariance of clock synchronization accuracy based on the evaluated time stamp observation model.

Optionally, the time-domain resource allocation optimization problem is expressed as:

P1:max{U(B,λ)};

C2: Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C3: 0<λ<1;

Wherein, P1 is the objective function of the time-domain resource allocation optimization problem, C1 to C3 are the constraints of the objective function P1, U(B,λ)=R _b , B is the length of the time-keeping band, and λ is the Packet arrival rate, R _b is the network throughput, P _k is the error covariance of the clock synchronization accuracy, E[P _k ] is the expected value of P _k , E[P _k ]=AXA ^T +Q-λAXC ^T ( CXC ^T +R) ^-1 CXA ^T , X=E[P _k-1 ], λ=E[γ _k ]=p _d , ΔE _k is the fixed error in the synchronization process, E[P _k-1 ] is P The expected value of _k-1 , E[γ _k ] is the expected value of γ _k , Tr(E[P _k ]) is the trace of E[P _k ], T _S is the length of the data packet time slot, A is the coefficient matrix, A ^T is the transposition matrix of coefficient matrix, Q is the covariance matrix of process noise, R is the covariance of observation noise, and P _k-1 is the error covariance matrix at time k-1.

Optionally, based on the above definition of the time stamp observation model, the following formula (3) is used to determine the error covariance of the clock synchronization accuracy:

P _k ＝E[e _k e _k ^T ] (3)

Wherein, P _k is the error covariance of the clock synchronization accuracy, _ek is the estimation error of the clock state variable, ^{ek T} _is the transposition matrix of _ek ,

x _k is the clock state variable,

is the estimation of x _k under the condition that the observed value is known,

Then, based on the Kalman filter algorithm, in the case of loss of time observation variable values, the following formulas (15) to (19) are used to update the error covariance P _k of the clock synchronization accuracy in each cycle of clock synchronization :

P _k ′＝AP _k A ^T +Q (16)

P _k+1 = P _k ′-γ _k+1 K _k+1 CP _k ′ (18)

K _k+1 ＝P _k ′C ^T (CP _k ′C ^T +R) ^-1 (19)

in,

is the x _k predicted by the Kalman filter algorithm, P _k ′ is the P _k predicted by the Kalman filter algorithm,

is the updated estimate of x _k+1 , γ _k+1 is a binary Bernoulli variable, K _k+1 is the Kalman filter gain of the k+1th cycle, z _k+1 is the k+1th cycle The clock observation variable value of .

Substituting formula (16) and formula (19) into formula (18), the recursive formula of P _k represented by the following formula (20) is obtained:

P _k+1 ＝AP _k A ^T +Q-γ _k AP _k C ^T (CP _k C ^T +R) ^-1 CP _k A ^T (20)

It can be seen from formula (20) that P _k is a random variable affected by γ _k ; therefore, it is very necessary to analyze P _k in a statistical sense. , obtain the recursive formula of E[P _k ] expressed in the following formula (21):

E[P _k ]＝AXA ^T +Q-λAXC ^T (CXC ^T +R) ^-1 CXA ^T (21)

Wherein, X=E[P _k-1 ], λ=E[γ _k ]=p _d .

Since high-precision timing services are sensitive to wireless communication delays, in order to prevent data packet collisions during data transmission caused by clock synchronization errors, in the 5G Orthogonal Frequency Division Multiple Access (OFDMA) time domain The time-guaranteed band is added to protect each transmission time slot, that is, a part is vacated between every two adjacent transmission time slots as a guard interval. Obviously, on the one hand, the longer the time-keeping band, the less data can be transmitted in each transmission slot; on the other hand, the too small time-keeping band will increase the probability of data collision, so the The length needs to be set reasonably, while taking into account the effectiveness and reliability of the communication. Here, the constraint conditions of the following formula (22) can be set for the length B of the time-guaranteed band:

B≥ΔE _k +{E[P _k ]} (22)

Among them, ΔE _k is the fixed error in the synchronization process, and {E[P _k ]} is the incremental synchronization variance. To simplify the calculation, the trace of E[P _k ] is used instead of {E[P _k ]}, that is, the following formula (23) represents the constraint condition of the time-guaranteed belt length B:

B≥ΔE _k +Tr(E[P _k ]) (23)

Based on the above analysis of deterministic networks with clock synchronization accuracy requirements, it is necessary to ensure that the overhead of time-domain resources is balanced and the network throughput is maximized under the constraints of clock synchronization accuracy, so this problem is determined as a time-domain Resource allocation optimization problem, the resource allocation optimization problem in time domain is expressed as:

P1:max{U(B,λ)};

C2: Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C3: 0<λ<1.

Among them, U(B,λ)=R _b , constraint condition C1 indicates that the sequence of E[P _k ] tends to be stable under long-term conditions, that is, to ensure the accuracy of clock synchronization; constraint condition C2 indicates the design criterion of the time-guarantee band length, That is, it is not less than the critical value that causes data to collide, and at the same time cannot exceed the length of the transmission time slot; the constraint condition C3 indicates that the arrival rate of data packets is limited between (0,1).

Further, converting the time-domain resource allocation optimization problem into a convex optimization problem: solving the convex optimization problem to obtain the data packet arrival rate.

P2:min{Tr(P _k )E[a _k |P _k ]-VU}

stC4:Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C5:0<λ<1;

P _k+1 = max{P _k -b _k ,0}+a _k , b _k is the departure process of the virtual queue, b _k is a constant, and b _k ≥ E[a _k ].

As an example, since the constraints of problem P1 contain nonlinear constraints, and it can be seen from the analysis that its feasible set and objective function are non-convex, it is difficult to solve P1 directly. The present invention adopts Lyapunov ( Lyapunov) optimization method to analyze and solve the problem P1.

First, based on the Lyapunov optimization theory, the problem P1 is transformed into a joint optimization problem of error covariance and throughput of clock synchronization accuracy. Specifically, for the error covariance P _k , the recursive definition based on formula (20) adopts the virtual queue represented by the following formula (24):

P _k+1 ＝max{P _k -b _k ,0}+a _k (24)

Among them, P _k+1 is the virtual queue, and b _k is the departure process of the virtual queue. For the convenience of calculation, it is assumed that b _k is a constant and satisfies b _k ≥ E[a _k ] to ensure the stability of the queue. E[a _k ] is the expected value of a _k ; a _k is the random arrival process of the virtual queue, its randomness is introduced by the random variable γ _k in (20), then a _k can be calculated by the formula (20), and a _k uses the following Formula (25) expresses:

Based on the definition of the virtual queue in formula (24), the Lyapunov function L(P _k ) about P _k can be expressed by the following formula (26):

L(P _k )=P _k ² /2 (26)

Then the Lyapunov drift function Δ(P _k ) is expressed by the following formula (27):

Δ(P _k )=L(P _k+1 )-L(P _k )≤D+P _k E[a _k |P _k ] (27)

Among them, L(P _k+1 ) is the Lyapunov function about P _k+1 , D is the boundary value, D=(E[a _k ² |P _k ]+b _k ² )/2, E[a _k ² | P _k ] is the expected value of the second moment of the random variable a _k under the condition that P _k is known.

Then the drift plus penalty function is expressed by the following formula (28):

Δ(P _k )-VU≤D+P _k E[a _k |P _k ]-VU (28)

Among them, V is the control coefficient, and V is used to adjust the impact of the drift plus penalty function on the overall, so that the problem P1 can be transformed into a Lyapunov optimization problem, that is, a convex optimization problem represented by the problem P2:

P2: min{Tr(P _k )E[a _k |P _k ]-VU}

stC4:Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C5: 0<λ<1.

Furthermore, it can be seen from problem P2 that although the constraints of problem P2 have been simplified, since the constraints in C4 and C5 are not independent, that is, the change of λ will affect the value of E[P _k ] , and then affect the constraints of B. But it is not difficult to see that the change of the length B of the time-guaranteed band only directly affects U in the objective function of the problem P2, and U is a monotonically decreasing function about B, so the value of B only needs to be set to the minimum value that satisfies the constraint condition C2 value, that is, B=Tr(E[P _k ])+ΔE _k . Then problem P2 can be simplified as:

P3: min{Tr(P _k )E[a _k |P _k ]-VU}

s.t.C6: 0<λ<1

Among them, P3 is the objective function of the convex optimization problem, and C6 is the constraint condition of the objective function P3.

For example, in the case of B=Tr(E[P _k ])+ΔE _k , the problem P3 is solved to obtain the data packet arrival rate.

The present invention solves the problem P3 based on an improved algorithm of deep Q-learning (DQN), that is, Dueling-DDQN (hereinafter referred to as 3DQN). The 3DQN algorithm is mainly composed of three parts: state, action and reward, in which the state space S={P _k ,U,B,λ} reflects the value of each parameter of the environment in the current step in the 3DQN algorithm; the action space a={ x _inc , x _de }, the value of a {1,0} means that λ increases according to the step size, and the value of a {0,1} means that λ decreases according to the step size; the reward function r=Lya _pre -Lya _cur , where, Lya _pre indicates the value of the objective function of the problem P3 before each iteration of the algorithm is updated, and Lya _cur indicates the value of the objective function of the problem P3 after each iteration of the algorithm, that is, r reflects the reduction of the objective function of the algorithm in one iteration When the algorithm obtains the optimal solution, the value of r tends to be stable, and the system converges to the optimal state.

Specifically, in the case of B=Tr(E[P _k ])+ΔE _k , the process of solving problem P3 is as follows:

Obtain initial state variables and initial function values; wherein, the initial state variables include initial error covariance, initial data packet arrival rate, initial time zone length and initial network throughput;

Select the target action in the action space according to the initial state variables;

Execute the target action, calculate the cost reward of taking the target action under the initial state variable according to the reward function, and generate the next state variable;

Store each state variable, corresponding target action and cost reward in the experience pool;

When it is determined that the current number of iterations is greater than or equal to the preset threshold, data samples are obtained from the experience pool for training, and the evaluation value of the state is determined;

When the training process is executed for a preset number of times, the parameters of the current Q network are copied to the target Q network until the convergence condition is reached, and the optimal packet arrival rate is output.

The simulation results show that the algorithm proposed by the present invention has certain superiority in joint optimization of clock synchronization accuracy and throughput compared with existing strategies.

Optionally, step 104 in FIG. 1 may specifically be implemented in the following manner:

Determine the network throughput based on formula (4):

is the inverse function of Q(x),

1-λ is the probability of packet loss, that is, the transmission error rate, and t is the integral variable.

For example, in the deterministic communication scenario of the power grid, the impact of reliability on network performance is greater than that of the network in general scenarios, and the throughput is mainly reflected in the impact of the error rate of data packet transmission on it. The present invention considers the transmission The error rate correction channel capacity calculation method is used to obtain high reliability network throughput, and the network throughput calculation is expressed by the above formula (4). When determining the length of the time-keeping band B and calculating the optimal data packet arrival rate λ, the length of the time-keeping band B and the data packet arrival rate λ are brought into formula (4) to obtain the maximum network throughput.

The apparatus for configuring time-domain resources provided by the present invention is described below, and the apparatus for configuring time-domain resources described below and the method for configuring time-domain resources described above may refer to each other correspondingly.

Fig. 4 is a schematic structural diagram of a time-domain resource configuration device provided by the present invention. As shown in Fig. 4, the time-domain resource configuration device includes an acquisition unit 4014, a construction unit 402, a solving unit 4034, and a determination unit 404; wherein:

An acquisition unit 401, configured to acquire the error covariance of the clock synchronization accuracy;

The construction unit 402 is configured to construct a time-domain resource allocation optimization problem based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length, and the arrival rate of data packets; wherein, the time-domain resource allocation optimization problem is to ensure the clock synchronization accuracy The problem of maximizing network throughput under the condition of ;

A solving unit 403, configured to solve the time-domain resource allocation optimization problem to obtain the data packet arrival rate;

A determining unit 404, configured to determine the network throughput based on the data packet arrival rate and the time-keeping band length.

A time-domain resource configuration device provided by the present invention constructs a time-domain resource configuration optimization problem based on the error covariance of clock synchronization accuracy, the length of the time-guaranteed band, and the arrival rate of data packets, and solves the problem of time-domain resource allocation based on the length of the time-guaranteed band and the The packet arrival rate calculation obtained from the configuration optimization problem maximizes the network throughput. Since the maximized network throughput reflects the allocation of time domain resources, the optimization of time domain resource allocation is realized.

Based on any of the above embodiments, the acquiring unit 401 is specifically configured to:

Evaluating network delay certainty based on the time stamp observation model;

Based on any of the above-mentioned embodiments, the acquiring unit 401 is further specifically configured to:

Build the time stamp observation model based on formula (1);

z _i,k =γ _k [T _i,1 +T _i,2 -2t]=γ _k [Cx _i (k)+v _i (k)] (1)

p _d ＝p{t _l <x<αt _u }p{0<w<(1-α)t _u } (2)

θ satisfies the constraint condition E[e ^θA(1) ]E[e ^-θS(1) ]≤1, A(1) is the amount of data arrival per unit time, S(1) is the data service volume per unit time, S[(1-α)t _u ] is the data service volume in (0,(1-α)t _u ) period;

P _k ＝E[e _k e _k ^T ] (3)

x _k is the clock state variable,

is the estimation of x _k under the condition that the observed value is known,

Based on any of the above-mentioned embodiments, the time-domain resource allocation optimization problem is expressed as:

P1:max{U(B,λ)};

C2: Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C3: 0<λ<1;

Based on any of the above embodiments, the solving unit 403 is specifically configured to:

P2:min{Tr(P _k )E[a _k |P _k ]-VU}

stC4:Tr(E[P _k ])+ΔE _k ≤ B<T _S ;

C5:0<λ<1;

Solving the convex optimization problem to obtain the data packet arrival rate.

Based on any of the above embodiments, the determining unit 404 is specifically configured to:

Determine the network throughput based on formula (4);

is the inverse function of Q(x),

1-λ is the probability of packet loss, and t is the integral variable.

Fig. 5 is a schematic diagram of the physical structure of the electronic device provided by the present invention. As shown in Fig. 5, the electronic device may include: a processor (processor) 510, a communication interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540 , wherein, the processor 510 , the communication interface 520 , and the memory 530 communicate with each other through the communication bus 540 . The processor 510 can call the logic instructions in the memory 530 to execute the time domain resource configuration method, the method includes: obtaining the error covariance of the clock synchronization accuracy;

In addition, the above logic instructions in the memory 530 may be implemented in the form of software function units and be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disc, etc., which can store program codes. .

On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can Executing the time-domain resource allocation method provided by the above methods, the method includes: obtaining the error covariance of the clock synchronization accuracy;

In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it is implemented to execute the time-domain resource allocation method provided by the above-mentioned methods, the method Including: obtaining the error covariance of clock synchronization accuracy;

The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative efforts.

Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic discs, optical discs, etc., including several instructions to make a computer device (which may be a personal computer, server, or network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

A time-domain resource allocation method, characterized in that it includes:

Obtain the error covariance of clock synchronization accuracy;

Based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length and the arrival rate of data packets, the time-domain resource allocation optimization problem is constructed; wherein, the time-domain resource allocation optimization problem is to make the network throughput under the condition of ensuring the clock synchronization accuracy The problem of maximizing the quantity;

Solving the time-domain resource allocation optimization problem to obtain the data packet arrival rate;

The network throughput is determined based on the data packet arrival rate and the length of the timekeeping band.
The method for configuring time-domain resources according to claim 1, wherein the error covariance of obtaining clock synchronization accuracy comprises:

Construct a timestamp observation model based on timestamp information and clock state variables;

Evaluating network delay certainty based on the time stamp observation model;

The error covariance of the clock synchronization accuracy is determined based on the evaluated timestamp observation model.
The time-domain resource configuration method according to claim 2, wherein said constructing a time stamp observation model based on time stamp information and clock state variables comprises:

Build the time stamp observation model based on formula (1);

z i,k =γ k [T i,1 +T i,2 -2t]=γ k [Cx i (k)+v i (k)] (1)

Among them, z i,k is the time stamp observation model, γ k is a binary Bernoulli random variable, i is the number of base station nodes, T i,1 is the time synchronization sent by base station node gNB i to timing terminal node UE j The moment of the variable, T i,2 is the moment when the base station node gNB i receives the reply time synchronization variable from the terminal node UE j , j is the number of terminal nodes, t is the ideal reference time, C is the observation matrix, C=[0 2] , x i (k) is the clock state variable of the base station node gNB i , k is the kth period of the clock synchronization process, v i (k) is the observation noise, v i (k)=d i +Y i,k , v i (k) is a Gaussian random variable that obeys the N(0,R) distribution, d i is the fixed time delay generated during the synchronization process, and Y i, k is the random time delay generated due to the randomness of the wireless channel.
The time-domain resource configuration method according to claim 3, wherein the evaluation of network delay certainty based on the time stamp observation model includes:

Solving the confidence degree of the wireless network delay certainty in the formula (2);

p d ＝p{t l <x<αt u }p{0<w<(1-α)t u } (2)

Among them, p d is the confidence degree of the wireless network delay determinism, α is the delay scaling factor, 0<α<t l /t u , t l is the lower bound of the sum of transmission delay and queuing delay, t u is the upper bound of the sum of transmission delay and queuing delay, x is the random variable of transmission delay, w is the random variable of queuing delay of nodes in the wireless network; p{t l <x<αt u } is the Probability value between t l and αt u ;
z(t)=(2 L/(Wt) -1)/c, L is the length of the transmitted data packet, W is the bandwidth of the channel, and c is the constant coefficient of the receiving SNR; p{0<w<(1 -α)t u } is the probability value of w falling between 0 and (1-α)t u ;
θ satisfies the constraint condition E[e θA(1) ]E[e -θS(1) ]≤1, A(1) is the amount of data arrival per unit time, S(1) is the data service volume per unit time, S[(1-α)t u ] is the data service volume in (0,(1-α)t u ) period;

The distribution of γ k is determined based on the confidence of the wireless network delay determinism.
The time-domain resource configuration method according to claim 4, wherein the error covariance of determining the clock synchronization accuracy based on the evaluated time stamp observation model comprises:

Determine the error covariance of the clock synchronization accuracy based on formula (3);

P k ＝E[e k e k T ] (3)

Wherein, P k is the error covariance of the clock synchronization accuracy, ek is the estimation error of the clock state variable, ek T is the transposition matrix of ek ,
x k is the clock state variable,
is the estimation of x k under the condition that the observed value is known,
The time-domain resource allocation method according to claim 5, wherein the time-domain resource allocation optimization problem is expressed as:

P1:max{U(B,λ)};

C2: Tr(E[P k ])+ΔE k ≤ B<T S ;

C3:0<λ<1;

Wherein, P1 is the objective function of the time-domain resource allocation optimization problem, C1 to C3 are the constraints of the objective function P1, U(B,λ)=R b , B is the length of the time-keeping band, and λ is the Packet arrival rate, R b is the network throughput, P k is the error covariance of the clock synchronization accuracy, E[P k ] is the expected value of P k , E[P k ]=AXA T +Q-λAXC T ( CXC T +R) -1 CXA T , X=E[P k-1 ], λ=E[γ k ]=p d , ΔE k is the fixed error in the synchronization process, Tr(E[P k ]) is The trace of E[P k ], T S is the time slot length of the data packet, A is the coefficient matrix, A T is the transpose matrix of the coefficient matrix, Q is the covariance matrix of the process noise, R is the covariance of the observation noise, P k-1 is the error covariance matrix at time k-1.
The time domain resource allocation method according to claim 6, wherein said solving the time domain resource allocation optimization problem to obtain the data packet arrival rate comprises:

Convert the time-domain resource allocation optimization problem into a convex optimization problem:

P2:min{Tr(P k )E[a k |P k ]-VU}

stC4:Tr(E[P k ])+ΔE k ≤ B<T S ;

C5:0<λ<1;

Among them, P2 is the objective function of the convex optimization problem, C4 and C5 are the constraints of the objective function P2, Tr(P k ) is the trace of P k , a k is the random arrival process of the virtual queue, E[a k | P k ] is the expected value of a k when P k is known, V is the control coefficient,

P k+1 ＝max{P k -b k ,0}+a k , b k is the departure process of the virtual queue, b k is a constant, and b k ≥ E[a k ];

Solving the convex optimization problem to obtain the data packet arrival rate.
The method for configuring time-domain resources according to claim 7, wherein said determining said network throughput based on said data packet arrival rate and said time-guaranteed band length comprises:

Determine the network throughput based on formula (4);

Among them, W e is the equivalent bandwidth of the time-guaranteed band length,
is the inverse function of Q(x),
1-λ is the probability of packet loss, and t is the integral variable.
A time-domain resource allocation device, characterized in that,

An acquisition unit, configured to acquire the error covariance of the clock synchronization accuracy;

A construction unit for constructing a time-domain resource configuration optimization problem based on the error covariance of the clock synchronization accuracy, the time-guaranteed band length, and the arrival rate of data packets; wherein, the time-domain resource configuration optimization problem is to ensure clock synchronization accuracy The problem of maximizing network throughput under the condition;

A solving unit, configured to solve the time-domain resource allocation optimization problem to obtain the data packet arrival rate;

A determining unit, configured to determine the network throughput based on the arrival rate of the data packet and the length of the time guarantee band.
An electronic device, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, characterized in that, when the processor executes the program, it implements claims 1 to 8 Steps in any one of the time-domain resource configuration methods.
A non-transitory computer-readable storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method for configuring time-domain resources according to any one of claims 1 to 8 are implemented .
A computer program product, comprising a computer program, characterized in that, when the computer program is executed by a processor, the steps of the method for configuring time-domain resources according to any one of claims 1 to 8 are implemented.