A method for optimizing the latency of LTE systems in unlicensed frequency bands

Abstract

The present invention discloses an optimization method for the delay of an unlicensed frequency band LTE system, including: obtaining the state changes of LTE-U nodes and WiFi nodes in the system based on a pre-established system model; obtaining the delay under the conflict of real-time service LTE-U nodes according to the state changes; obtaining the probability that all real-time service LTE-U nodes do not exceed the delay threshold based on a given delay threshold; summing the probabilities that all real-time service LTE-U nodes do not exceed the delay threshold to obtain the total probability value of all real-time service LTE-U nodes, and obtaining the range of delay sensitivity coefficients according to the given constraint probability; adjusting the upper limit and lower limit of the contention window generated by the backoff algorithm in real time according to the range of delay sensitivity coefficients. The present invention introduces a delay influencing factor according to the delay of the real-time service node, introduces the contention window generated by the backoff algorithm, thereby adjusting the upper limit and lower limit of the contention window in real time, and optimizing the delay accuracy of the real-time service in the LTE system.

Description

Technical Field

[0001] The present invention belongs to the field of communication technology, and in particular relates to a method for optimizing the time delay of an unlicensed frequency band LTE system. Background Art

[0002] The vigorous development of mobile Internet has brought about a rapid growth of smart devices and mobile applications. Faced with such a huge amount of device access and business volume, the industry and academia have proposed the concept of deploying LTE in unlicensed bands, namely LTE-U technology. LTE-U is a technology that uses the LTE protocol standard to communicate on the WiFi band and aggregates the licensed and unlicensed bands through carrier aggregation (CA). Since LTE uniformly schedules wireless resources through base stations, after introducing LTE into the free and public WiFi band, it will inevitably compete with the WiFi system in the same band. In order to ensure the coexistence of LTE and WiFi in the unlicensed band, the 3GPP organization proposed the Licensed Assisted Access (LAA) technology. The core of LAA is to adopt the Listen Before Talk (LBT) mechanism, that is, before the device accesses the channel, it first undergoes channel detection, but this listen-before-send method will inevitably bring delays to the LTE system in the unlicensed band.

[0003] Regarding the performance optimization of LTE-U and WiFi coexistence systems, existing technical research focuses on improving the throughput of LTE-U systems. However, LTE-U services are different from those of WiFi networks. In LTE-U, there are real-time services that are sensitive to latency, such as voice and video services. If only the throughput of the system is improved, the latency requirements of such services cannot be met. Summary of the invention

[0004] Purpose of the invention: In order to overcome the deficiencies in the prior art, the present invention provides a method for optimizing the latency of an unlicensed band LTE system, divides the service types within LTE-U, introduces a latency influencing factor into the latency of a real-time service node, and introduces a contention window generated by a backoff algorithm, thereby adjusting the upper and lower limits of the contention window in real time, and optimizing the latency accuracy of real-time services in the LTE system; and solves the problem of excessive latency of the LTE system after LTE adopts the LBT mechanism to access the WiFi band.

[0005] Technical solution: In the first aspect, the present invention provides a method for optimizing the latency of an unlicensed band LTE system, comprising:

[0006] Based on a pre-established system model of a coexistence scenario of an LTE-U system and a WiFi system, the LTE-U nodes and the WiFi nodes in the coexistence scenario system are analyzed through a two-dimensional Markov state transition diagram to obtain state changes of the LTE-U nodes and the WiFi nodes in the system; wherein the system model of the coexistence scenario of the LTE-U system and the WiFi system is established according to a mechanism of LTE accessing a WiFi frequency band;

[0007] According to the status changes of LTE-U nodes and WiFi nodes in the system, the delay of real-time service under LTE-U node conflict is obtained;

[0008] According to the delay under the conflict of real-time service LTE-U nodes and the predetermined delay threshold, the probability that all real-time service LTE-U nodes do not exceed the delay threshold is calculated, and the probability that all real-time service LTE-U nodes do not exceed the delay threshold is summed up to obtain the total probability value of all real-time service LTE-U nodes in the system;

[0009] The total probability value of all real-time service LTE-U nodes is compared with the given constraint probability, and the value range of the delay sensitivity coefficient is obtained according to the total probability value less than the constraint probability;

[0010] The upper limit and lower limit of the contention window generated by the backoff algorithm are adjusted in real time according to the value range of the delay sensitivity coefficient, so as to optimize the delay accuracy of the real-time service in the LTE system.

[0011] In a further embodiment, the LTE-U system services also include common services.

[0012] In a further embodiment, obtaining the delay of the real-time service LTE-U node under the conflict includes:

[0013] According to the state changes of LTE-U nodes and WiFi nodes in the system, the single node transmission probability and collision probability of LTE-U nodes and WiFi nodes are obtained respectively;

[0014] From the transmission probability of a single LTE-U node and the collision probability of a single node, the real-time service LTE-U node and the common service LTE-U node are determined, and the real-time service LTE-U node is calculated to obtain the delay under the conflict of the real-time service LTE-U node;

[0015] Among them, the state change transfer equation is expressed as:

[0016]

[0017] Where i represents the backoff order, k represents the backoff counter value at the current order, P(i,k) represents the probability when the state is (i,k) on the two-dimensional Markov chain, W0 represents the initial backoff window length, and W i represents the backoff window length when the backoff order is i, W w represents the value of the initial backoff window, P w Represents the probability of node collision.

[0018] In a further embodiment, the transmission probability calculation process of a single node is:

[0019] Substitute the limit steady-state distribution probability formula of the Markov chain into the state change transfer equation to obtain the steady-state probability when the backoff value is 0;

[0020] Combined with the regularity of Markov chain, the steady-state probability of LTE-U node in any state during the backoff process is obtained;

[0021] The steady-state probability in any state during the backoff process is classified and discussed by backoff order i, and a simplified formula for the steady-state probability of LTE-U nodes in any state is obtained.

[0022] When the sum of the steady-state distribution probabilities is 1, the transmission probability of a single node can be obtained by substituting the simplified formula of the steady-state probability under any state of the LTE-U node into the formula;

[0023] Among them, the limit steady-state distribution probability is defined as:

[0024]

[0025] Where b i,k represents the limit steady-state distribution probability, s(t) represents the backoff order of the node at time t, and b(t) represents the value of the backoff counter of the node at time t;

[0026] The calculation formula of the steady-state probability of the LTE-U node during the backoff process is:

[0027]

[0028] Where b i,0 represents the steady-state probability when the state is (i,0), b m,0 represents the steady-state probability when the state is (m,0), b 0,0 represents the steady-state probability at the initial state (i, 0), and M represents the lower limit of the backoff order i;

[0029] Combined with the regularity of Markov chain, when k∈[0,W i -1], the calculation formula of the steady-state probability of the LTE-U node in any state is:

[0030]

[0031]

[0032] Where, formula (4) to formula (5) b j,0 It represents the steady-state probability when the backoff order is j and the backoff value is 0;

[0033] When the backoff order is 0<i<M:

[0034]

[0035] When the backoff order i = M:

[0036]

[0037] Combining formulas (5)-(7), formula (4) is simplified to:

[0038]

[0039] When the sum of the steady-state distribution probabilities is 1, the simplified formula for the steady-state probability of the LTE-U node in any state is expressed as:

[0040]

[0041] Formula (9) is solved as follows:

[0042]

[0043] The transmission probability of a single node is:

[0044]

[0045] In the formula, τ w represents the transmission probability of a single node, represents the collision probability at backoff order i, The collision probability when the backoff order is maximum.

[0046] In a further embodiment, the collision probability of a single node is calculated by calculating the transmission probability τ of a single node w It is inferred that the inference calculation formula is as follows:

[0047] P w =1-(1-τ w ) n (12)

[0048] Where n is the number of nodes participating in the competition in the system.

[0049] In a further embodiment, a method for calculating the probability that all real-time service LTE-U nodes do not exceed the delay threshold includes:

[0050] According to the state changes of LTE-U nodes and WiFi nodes in the system, the state selection probability and conditional probability of a single node are calculated respectively;

[0051] The state selection probability and conditional probability of a single node are multiplied to obtain the probability that a single real-time service LTE-U node transmits successfully in state (i, k) and the delay does not exceed the delay threshold Dth;

[0052] The real-time service LTE-U node is in state (i, k) and the transmission is successful and the delay does not exceed the delay threshold D th The probability calculation formula is:

[0053] P(D th ; i, k) = Q i [k]·Φ(D th |(i,k)) (13)

[0054] In the formula, P(D th ; i, k) indicates that the real-time service LTE-U node is successfully transmitted in state (i, k) and the delay does not exceed the delay threshold D th The probability of Q i [k] represents the state selection probability of the real-time service node data packet being successfully transmitted in the i-th order and k-th backoff time slot, Φ(D th |(i,k)) indicates that the real-time service node successfully transmits at the i-th order and k-th backoff slot, and the delay does not exceed D th The conditional probability of .

[0055] In a further embodiment, the state selection probability is calculated as follows:

[0056] According to the status changes of LTE-U nodes and WiFi nodes in the system, real-time service nodes are screened, and the transmission probability of real-time service nodes is calculated to obtain the collision probability of a single node;

[0057] Substituting the collision probability of a single node into the conditional probability calculation, we can get the probability of successful transmission of a single data packet;

[0058] The probability of successful transmission of a single data packet is multiplied by the probability density function of successful transmission of the data packet in state (i, k) to obtain the state selection probability of a single node;

[0059] The calculation formula for state selection probability is:

[0060] Q i [k] = W i-s,k ·H i,k (14)

[0061] In the formula, Q i [k] represents the probability of successful transmission of the real-time service node data packet in the i-th order and k-th backoff time slot, W i-s,k represents the probability of successful transmission of a single data packet, H i,k The probability density function of the successful transmission of a data packet in state (i, k);

[0062] The formula for calculating the probability of successful transmission of a single data packet is:

[0063]

[0064] Where Pl-s is the collision probability of real-time service nodes; S l is the maximum backoff order, i is the backoff order, and k is the number of time slots under the current backoff order;

[0065] The probability density function expression of the successful transmission of a data packet in state (i, k) is:

[0066] H i,k =(h0*h1*···*h i )[k] (16)

[0067] In the formula, (h0*h1*···*h i )[k] represents h0[k],h1[k],...,h i Discrete convolution of [k].

[0068] In a further embodiment, the conditional probability is calculated as follows:

[0069] Combined with the two-dimensional Markov state transition diagram, the probability of channel idleness, the probability of successful transmission of a single WiFi node, the probability of successful transmission of a non-real-time service node, the probability of conflict between nodes, the data packet length of the WiFi site, the data packet length transmitted by the real-time LTE-U node, and the average duration of the time slot when the other sites collided when the real-time LTE-U node was in the backoff state, the correlation formula was derived;

[0070] According to the state changes of LTE-U nodes and WiFi nodes in the system, the average duration of the time slot when the real-time service node is in the backoff state and the average time slot time when the real-time service node participates in the transmission are obtained; and the average duration of the time slot when the real-time service node is in the backoff state and the collision between the nodes are substituted into the correlation formula to obtain the mean value E of the length of a single backoff time slot when the real-time service node does not participate in the transmission. l and the variance D of real-time service nodes not participating in transmission l ;

[0071] By calculating the mean value E of a single backoff slot length l , the variance D of real-time service nodes not participating in transmission l When the real-time service nodes participate in the transmission, the average time slot time of the collision between nodes is substituted into the solution formula of the expectation and variance of a single node to obtain the expectation E of the real-time service node in state (i, k) i,k and variance D i,k ;

[0072] The expected value E of a real-time service node in state (i, k) i,k and variance D i,k Combined with the central limit theorem, the delay is determined And delay and the expectation E of the corresponding state (i, k) i,k and variance D i,k Substituting into the calculation formula of conditional probability, we can get the condition that the real-time service node successfully transmits in the i-th order and k-th backoff time slot, and the delay does not exceed D th The conditional probability of

[0073] The derived correlation formulas among the channel idle probability, the probability of successful transmission of a single WiFi node, the probability of successful transmission of a non-real-time service node, and the probability of conflict between nodes are as follows:

[0074]

[0075]

[0076]

[0077] P c =1-P idle -P w -P l-s (17)

[0078] In the formula, τ w represents the transmission probability of the WiFi node, n w represents the number of WiFi nodes; τ l-o represents the transmission probability of a common service node, P idle represents the channel idle probability, P w represents the probability of a WiFi node successfully transmitting, P l - s represents the probability of successful transmission of non-real-time service nodes, P c represents the probability of conflict between nodes;

[0079] When the real-time service node is in the backoff state, the average duration of the time slot when the other stations collide is T cThe expression is:

[0080]

[0081] The average length of a single backoff slot when the real-time service node does not participate in the transmission E l and the variance D of real-time service nodes not participating in transmission l The calculation formula is:

[0082]

[0083] Substituting formulas (17) and (18) into (19), we can calculate the mean value E of a single backoff slot length: l and variance D l ;P idle represents the channel idle probability, P w represents the probability of a WiFi node successfully transmitting, P l-s represents the probability of successful transmission of non-real-time service nodes, P c represents the probability of conflict between nodes, T w Indicates the data packet length of the WiFi site; T l-s Indicates the length of the data packet transmitted by real-time LTE-U; T c represents the average duration of the time slot when the other sites collide when the real-time service LTE-U node is in the backoff state, δ represents the length of the idle time slot, and E l and D l They represent the mean of the time slots in which the real-time service node does not participate in the transmission and the variance of the time slots in which the real-time service node does not participate in the transmission respectively;

[0084] When real-time service nodes participate in transmission, the average time slot time T of collision between nodes is c The solution formula is:

[0085]

[0086] The expected value E of a real-time service node in state (i, k) i,k and variance D i,k The solution formula is:

[0087]

[0088] In the formula, the mean value E of the length of a single backoff slot in formula (19) is l and variance D l and T in formula (20) c Substituting into formula (21), we can get the expected E of the real-time service node in state (i, k): i,k and variance D i,k , Tc represents the average time slot of collision between nodes when real-time service nodes participate in transmission; T sl Indicates the average length of a data packet transmitted by a real-time service node;

[0089] According to the central limit theorem, the delay is determined The expression of Gaussian distribution is:

[0090]

[0091] The formula for calculating conditional probability is:

[0092]

[0093] In the formula, Φ(D th |(i,k)) indicates that the real-time service node successfully transmits at the i-th order and k-th backoff slot, and the delay does not exceed D th The probability of represents the MAC delay of a real-time service node successfully transmitting a data packet at the i-th order and k-th time slot, E i,k for Expectations, D i,k for The variance of .

[0094] In a further embodiment, when the real-time service LTE-U node does not participate in the channel competition, the average duration of the time slot when the other stations collide is T′ c There are two situations:

[0095] Case 1: At least one WiFi node and a common LTE-U node transmit simultaneously in the same time slot and a collision occurs. The corresponding collision duration is max{T w ,T l-o}, the calculation expression of collision probability is:

[0096]

[0097] Case 2: Two or more WiFi nodes transmit at the same time in the same time slot and a collision occurs, and the LTE-U node does not attempt to transmit. The corresponding collision duration is T w , the calculation expression of collision probability is:

[0098]

[0099] In a further embodiment, when the real-time service LTE-U node participates in the channel competition, another situation is also included, and the corresponding collision duration is T l , the collision probability calculation expression is:

[0100]

[0101] In a further embodiment, a method for obtaining a value range of a delay sensitivity coefficient according to a total probability value less than a constraint probability includes:

[0102] The expectation E of the real-time business node in state (i, k) i,k and variance D i,k Substituting the conditional probability, we can conclude that when the real-time service node successfully transmits state (i, k), the delay does not exceed D th The probability Φ(D th |(i,k)) value, and then deduce that the real-time service node is in state (i,k) and the transmission is successful and the delay does not exceed D th The probability P(D th ; i, k);

[0103] The delay of successful transmission of real-time service LTE-U node in all states is no more than D th The probability P(D th ; the sum of i,k)P all Compare with the given constraint probability and select P(D th ; i, k);

[0104] For all the screened P(D th ; i, k) The delay sensitivity coefficient μ is solved to obtain the value range of the delay sensitivity coefficient;

[0105] Among them, for all the screened P(D th ; i, k) are used to derive the formula:

[0106]

[0107] In the formula, F[·] -1 represents the inverse Fourier transform; μ represents the delay sensitivity coefficient, S l represents the maximum backoff order, W l represents the initial backoff window, P l-s represents the collision probability of real-time business nodes, represents a random variable, E i,k and D i,k Denote random variables The expectation and variance of

[0108] P l-s , E i,k and D i,k The transmission probability τ of the real-time service node l-s Regarding l-sDepends on the delay sensitivity coefficient μ, and then derives the delay sensitivity coefficient μ, and the expression of the delay sensitivity coefficient μ is as follows:

[0109]

[0110] The delay of successful transmission of real-time service LTE-U node in all states is no more than D th The probability P(D th ; the sum of i,k)P all The expression is:

[0111]

[0112] In the formula, P(D th ; i, k) is the real-time LTE-U node successfully transmitting in state (i, k) with a delay no greater than D th The probability of i represents the backoff order, k represents the number of time slots experienced under the current backoff order, S l is the lower limit of the backoff order, W l-s is the contention window of the real-time LTE-U node;

[0113] Screen out P(D th ; The expression of i,k) is:

[0114]

[0115] In the formula, Po ut represents the given constraint probability, represents the collision probability of real-time LTE-U nodes under any backoff order, It represents the collision probability of real-time LTE-U node under the maximum backoff order.

[0116] In a further embodiment, the method for adjusting the upper limit and the lower limit of the contention window generated by the backoff algorithm in real time includes:

[0117] The delay sensitivity coefficient is introduced into the binary exponential backoff algorithm, so that the binary exponential backoff algorithm can adjust the contention window based on the delay sensitivity;

[0118] The formula for adjusting the contention window is:

[0119]

[0120] Where μ is the delay sensitivity coefficient, m l is the maximum backoff order, W l is the size of the initial contention window.

[0121] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0122] 1. Divide the service types within LTE-U, introduce delay influencing factors to the delay of real-time service nodes, introduce the contention window generated by the backoff algorithm, so as to adjust the upper and lower limits of the contention window in real time and optimize the delay accuracy of real-time services in the LTE system.

[0123] 2. During the analysis of the MAC delay of the LTE-U system, the duration of the idle time slot and the busy time slot of the channel are distinguished, which improves the accuracy of the delay evaluation of the LTE-U system.

[0124] 3. By dynamically adjusting the size of the real-time service contention window, the latency requirements of real-time services are guaranteed, thereby effectively improving the MAC latency of the LTE-U system. BRIEF DESCRIPTION OF THE DRAWINGS

[0125] Figure 1 It is a schematic diagram of the CSMA / CA algorithm in the basic access mode;

[0126] Figure 2 A schematic diagram of the process of LTE accessing the WiFi system;

[0127] Figure 3 A communication scenario problem model diagram established for a coexistence system of LTE using a WiFi frequency band according to the present invention;

[0128] Figure 4 Schematic diagram of Markov state transition of nodes in the coexistence system;

[0129] Figure 5 A schematic diagram of an improved real-time service contention window generation algorithm according to the present invention;

[0130] Figure 6 A schematic diagram of the process of calculating the probability of MAC delay threshold for real-time service LTE-U nodes;

[0131] Figure 7 is a distribution diagram of the probability that the real-time LTE-U meets the conditions;

[0132] Figure 8 Schematic diagram of the LTE-U latency optimization algorithm proposed in the present invention. DETAILED DESCRIPTION

[0133] In order to more fully understand the technical content of the present invention, the technical solution of the present invention is further introduced and illustrated in conjunction with specific embodiments below, but is not limited thereto.

[0134] The present invention provides a method for optimizing the time delay of an unlicensed frequency band LTE system, comprising:

[0135] Based on a pre-established system model of a coexistence scenario of an LTE-U system and a WiFi system, the LTE-U nodes and the WiFi nodes in the coexistence scenario system are analyzed through a two-dimensional Markov state transition diagram to obtain state changes of the LTE-U nodes and the WiFi nodes in the system; wherein the system model of the coexistence scenario of the LTE-U system and the WiFi system is established according to a mechanism of LTE accessing a WiFi frequency band;

[0136] According to the status changes of LTE-U nodes and WiFi nodes in the system, the delay of real-time service under LTE-U node conflict is obtained;

[0137] According to the delay under the conflict of real-time service LTE-U nodes and the predetermined delay threshold, the probability that all real-time service LTE-U nodes do not exceed the delay threshold is calculated, and the probability that all real-time service LTE-U nodes do not exceed the delay threshold is summed up to obtain the total probability value of all real-time service LTE-U nodes in the system;

[0138] The total probability value of all real-time service LTE-U nodes is compared with the given constraint probability, and the value range of the delay sensitivity coefficient is obtained according to the total probability value less than the constraint probability;

[0139] The upper limit and lower limit of the contention window generated by the backoff algorithm are adjusted in real time according to the value range of the delay sensitivity coefficient, so as to optimize the delay accuracy of the real-time service in the LTE system.

[0140] LTE-U system services also include common services.

[0141] The delay of real-time service LTE-U node conflict includes:

[0142] According to the state changes of LTE-U nodes and WiFi nodes in the system, the single node transmission probability and collision probability of LTE-U nodes and WiFi nodes are obtained respectively;

[0143] From the transmission probability of a single LTE-U node and the collision probability of a single node, the real-time service LTE-U node and the common service LTE-U node are determined, and the real-time service LTE-U node is calculated to obtain the delay under the conflict of the real-time service LTE-U node;

[0144] Furthermore, the transmission probability calculation process of a single node is:

[0145] Substitute the limit steady-state distribution probability formula of the Markov chain into the state change transfer equation to obtain the steady-state probability when the backoff value is 0;

[0146] Combined with the regularity of Markov chain, the steady-state probability of LTE-U node in any state during the backoff process is obtained;

[0147] The steady-state probability in any state during the backoff process is classified and discussed by backoff order i, and a simplified formula for the steady-state probability of LTE-U nodes in any state is obtained.

[0148] When the sum of the steady-state distribution probabilities is 1, the transmission probability of a single node can be obtained by substituting the simplified formula of the steady-state probability under any state of the LTE-U node into the formula;

[0149] The calculation method of the probability that all real-time service LTE-U nodes do not exceed the delay threshold includes:

[0150] According to the state changes of LTE-U nodes and WiFi nodes in the system, the state selection probability and conditional probability of a single node are calculated respectively;

[0151] The state selection probability and conditional probability of a single node are multiplied to obtain the success of a single real-time service LTE-U node transmission in state (i, k) and the delay does not exceed the delay threshold D th probability;

[0152] Furthermore, the calculation process of state selection probability is as follows:

[0153] According to the status changes of LTE-U nodes and WiFi nodes in the system, real-time service nodes are screened, and the transmission probability of real-time service nodes is calculated to obtain the collision probability of a single node;

[0154] Substituting the collision probability of a single node into the conditional probability calculation, we can get the probability of successful transmission of a single data packet;

[0155] The probability of successful transmission of a single data packet is multiplied by the probability density function of successful transmission of the data packet in state (i, k) to obtain the state selection probability of a single node;

[0156] Furthermore, the conditional probability is calculated as follows:

[0157] Combined with the two-dimensional Markov state transition diagram, the probability of channel idleness, the probability of successful transmission of a single WiFi node, the probability of successful transmission of a non-real-time service node, the probability of conflict between nodes, the data packet length of the WiFi site, the data packet length transmitted by the real-time LTE-U node, and the average duration of the time slot when the other sites collided when the real-time LTE-U node was in the backoff state, the correlation formula was derived;

[0158] According to the state changes of LTE-U nodes and WiFi nodes in the system, the average duration of the time slot when the real-time service node is in the backoff state and the average time slot time when the real-time service node participates in the transmission are obtained; and the average duration of the time slot when the real-time service node is in the backoff state and the collision between the nodes are substituted into the correlation formula to obtain the mean value E of the length of a single backoff time slot when the real-time service node does not participate in the transmission. l and the variance D of real-time service nodes not participating in transmission l ;

[0159] By calculating the mean value E of a single backoff slot length l , the variance D of real-time service nodes not participating in transmission l When the real-time service nodes participate in the transmission, the average time slot time of the collision between nodes is substituted into the solution formula of the expectation and variance of a single node to obtain the expectation E of the real-time service node in state (i, k) i,k and variance D i,k ;

[0160] The expected value E of a real-time service node in state (i, k) i,k and variance D i,k Combined with the central limit theorem, the delay is determined And delay and the expectation E of the corresponding state (i, k) i,k and variance D i,k Substituting into the calculation formula of conditional probability, we can get the condition that the real-time service node successfully transmits in the i-th order and k-th backoff time slot, and the delay does not exceed D th The conditional probability of

[0161] Furthermore, when the real-time service LTE-U node does not participate in channel competition, the average duration of the time slot when the other stations collide is T′ c There are two situations:

[0162] Case 1: At least one WiFi node and a common LTE-U node transmit simultaneously in the same time slot and a collision occurs. The corresponding collision duration is max{T w ,T l-o};

[0163] Case 2: Two or more WiFi nodes transmit at the same time in the same time slot and a collision occurs, and the LTE-U node does not attempt to transmit. The corresponding collision duration is T w ;

[0164] Furthermore, when the real-time service LTE-U node participates in the channel competition, there is another situation, and the corresponding collision duration is T l;

[0165] Methods for obtaining the value range of the delay sensitivity coefficient according to the total probability value less than the constraint probability include:

[0166] The expectation E of the real-time business node in state (i, k) i,k and variance D i,k Substituting the conditional probability, we can conclude that when the real-time service node successfully transmits state (i, k), the delay does not exceed D th The probability Φ(D th |(i,k)) value, and then deduce that the real-time service node is in state (i,k) and the transmission is successful and the delay does not exceed D th The probability P(D th ; i, k);

[0167] The delay of successful transmission of real-time service LTE-U node in all states is no more than D th The probability P(D th ; the sum of i,k)P all Compare with the given constraint probability and select P(D th ; i, k);

[0168] For all the screened P(D th ; i, k) The delay sensitivity coefficient μ is solved to obtain the value range of the delay sensitivity coefficient;

[0169] The method of adjusting the upper limit and lower limit of the contention window generated by the backoff algorithm in real time includes:

[0170] A delay sensitivity coefficient is introduced into the binary exponential backoff algorithm, so that the binary exponential backoff algorithm can adjust the contention window based on the delay sensitivity.

[0171] Example 1

[0172] refer to Figure 1 Figure 1 is a schematic diagram of CSMA / CA in basic access mode. WiFi uses CSMA / CA (Carrier Sense Multiple Access / Collision Avoidance) algorithm to compete for channel resources. There are two modes of this algorithm to choose from. One is the basic access mode, such as Figure 1As shown, when two or more stations collide during transmission in the same time slot, a collision will occur, and the collision duration is one frame length. Another mechanism is the request to send / clear to send mode, that is, the sending station first sends an RTS frame, and only sends a data packet when it receives a CTS frame returned by the receiving end. Therefore, only when multiple stations try to transmit data packets will a conflict occur when sending the RTS frame, and the collision duration is equal to the duration of RTS. Generally speaking, the RTS / CTS mode is used to solve the hidden node problem in the channel. In the present invention, we adopt the basic access mode, that is, the collision time is equal to the frame length of the data packet.

[0173] refer to Figure 2 The figure is a flow chart of LTE accessing the WiFi band. LTE accesses the WiFi band through the Listen Before Talk (LBT) mechanism. All nodes, including WiFi, compete for transmission opportunities in the same channel. When multiple nodes access the same medium in the network, the backoff process can be used to reduce the probability of conflict between nodes. Figure 2 In the example, when node A finishes transmitting, after DIFS, other stations start their backoff process. When node B competes for the transmission opportunity and starts transmitting, other stations are in the remaining backoff process.

[0174] like Figure 3 As shown, the present invention considers the scenario where one LTE-U node and multiple WiFi nodes coexist in the same frequency band. Among them, the types of device communication within LTE-U are divided into real-time services and ordinary services. The number of real-time services is n l-s , the number of ordinary services is n l-o , n w The WiFi nodes and LTE-U nodes share the same unlicensed frequency band. Both the WiFi nodes and LTE-U nodes are in saturated traffic models, that is, there are always data packets to be sent in the queue.

[0175] like Figure 4 As shown, assuming that the channel is ideal, when the number of nodes is limited and in a saturated traffic state, the state change process of a single node can be represented by a two-dimensional Markov chain (i, k), where i represents the backoff order and k represents the backoff counter size at the current order.

[0176] For a single node, the state transition includes the following four cases: (1) When the node is in the i-th state, i∈[0,M], if it detects that the time slot is idle, the value of the backoff counter will be reduced by 1, and the state of the node will also be transferred from right to left with probability 1; (2) If the node successfully transmits a data packet at the i-th order, the state of the node will transfer from (i,0), i∈[0,M-1]) to (0,0), and then continue to monitor the channel to send the next data packet; (3) If the node collides (transmission is unsuccessful) at the i-1th order, the backoff order will be increased by 1, and the size of the backoff window will be increased from W to W. i-1 Changes to W i , the state of the node will transfer from (i-1,0) to (i,k), i∈[1,M],k∈[0,W i -1]; (4) When the backoff order i reaches the maximum backoff order M, if the data packet is still not transmitted successfully, the node will discard the data packet. When the data packet is transmitted successfully, the node will listen to the next data packet in the channel transmission queue again. Regardless of whether the transmission is successful or not, the backoff order M will become 0, and the size of the backoff counter will become k, k∈[0,W0-1], that is, the node transfers from state (M,0) to state (0,k), k∈[0,W0-1].

[0177] Combined with the Markov state transition model diagram, the above four situations can be expressed by the following state transition equations:

[0178]

[0179] make yes Figure 2 The related limit steady-state distribution probability, where s(t) represents the backoff order of the node at time t, and b(t) represents the value of the backoff counter of the node at time t, then according to the state transition equation, we can get:

[0180]

[0181] From this we get:

[0182]

[0183] According to the regularity of Markov chain, when k∈[0,W i -1], there are:

[0184]

[0185] Next, we simplify formula (4):

[0186] (1) When i = 0,

[0187]

[0188] (2) When 0<i<M,

[0189]

[0190] (3) When i = M,

[0191]

[0192] Combining equations (5)-(7), equation (4) can be simplified as:

[0193]

[0194] And because the sum of the steady-state distribution probabilities is 1,

[0195]

[0196] The solution is:

[0197]

[0198] Therefore, the transmission probability of a single node τ w for:

[0199]

[0200] Assuming the system is stable, the probability of a node collision is P w and the transmission probability τ w The relationship between can be expressed as:

[0201] P w =1-(1-τ w ) n (12)

[0202] Where n is the number of nodes participating in the competition in the system.

[0203] refer to Figure 5 The present invention achieves the low latency requirement of such services by changing the size of the contention window. Specifically, the contention window is dynamically generated by introducing a delay sensitivity coefficient μ in the binary exponential backoff algorithm. The coefficient represents the degree of service dependence on delay. Then, the backoff window size W of the delay-sensitive service is l-s,i It can be expressed as:

[0204]

[0205] Where μ is the delay sensitivity coefficient, m l is the maximum backoff order, W lis the size of the initial contention window.

[0206] refer to Figure 6 and Figure 7 refer to Figure 6 and Figure 7 They are respectively the MAC delay calculation flow chart of the LTE-U node and the probability distribution diagram of the involved conditional probabilities being met; according to the Markov state transition diagram of the LTE-U and WiFi coexistence system, it can be seen that the node needs to go through three processes to successfully transmit a data packet, namely, triggering the initial backoff counter, generating a new counter due to node conflict, and transmitting the data packet. The LTE-U delay we are discussing refers to the period of time from the moment the LTE-U node triggers the initial backoff counter to the completion of the data packet transmission. In order to distinguish the impact of service type on LTE-U delay, we divide the services within LTE-U into real-time services and ordinary services. At the same time, considering that ordinary services are less sensitive to delays, the present invention limits the delay of real-time services, thereby improving the MAC delay of the LTE-U system. Next, we analyze the delay of real-time services in detail.

[0207] The successful transmission of LTE-U data packets may occur in any state of the state transition diagram, so we first consider the probability P(D; i, k) that the real-time service node transmits successfully in state (i, k) with a delay not exceeding D, and then sum all states to obtain the probability that the MAC delay experienced by the successful transmission does not exceed D. Next, we first analyze the probability P(D; i, k) that the real-time service node transmits successfully in state (i, k) with a delay not exceeding D.

[0208] For the state (i, k) in the i-th order and the k-th backoff slot, let P(D; i, k) be the probability that the real-time service node transmits successfully in the state (i, k) and the delay does not exceed D. According to the conditional probability formula, we can decompose P(D; i, k) into Q i The product of [k] and Φ(D|(i,k)), that is:

[0209] P(D;i,k)=Q i [k]·Φ(D|(i,k)) (14)

[0210] Among them, the state selection probability Q i [k] is the probability that the real-time service node successfully transmits in the i-th order and k-th backoff slot, and the conditional probability Φ(D|(i,k)) is the probability that the MAC delay does not exceed D under the condition that the real-time service node data packet is successfully transmitted in the i-th order and k-th backoff slot. i [k] and Φ(D|(i,k)) are derived in detail.

[0211] (1) State selection probability Q i Analysis of [k]

[0212] According to the multiplication formula, the probability Q that a real-time service node data packet is successfully transmitted in the i-th order and k-th backoff time slot is i [k] is equal to the probability H that the node chooses state (i,k) i,k , and the probability of successful transmission in this state W i,k The product of .

[0213] First, we analyze the probability H of a node selecting state (i, k) i,k Let h i [k] is the probability of the real-time LTE-U node being in the i-th k-th backoff slot (at state (i, k)). According to the Markov state transition diagram, for the i-th state point, the probability of selecting any state in this hierarchy is the inverse of the window size. Therefore, the random variable h i The probability distribution of [k] is:

[0214]

[0215] Among them, W l-s It is the size of the i-th order contention window of the real-time service node.

[0216] Since real-time services have high requirements for latency, we need to reduce the latency of such services. The backoff window size W of the real-time service proposed in the present invention is l-s,i for:

[0217]

[0218] In the Markov state transfer process, for the i-th order state point (i, k), the initial state of the point is (0, W s ), where W s is the initial backoff value. With the conflict in the transmission process, the level increases continuously and finally reaches the i-th level. Therefore, when we analyze the probability of successful transmission of a node in state (i, k), we need to first find the probability distribution of the sum from the 0th level to the i-th level.

[0219] According to the distribution properties of random variables and in probability theory, if X and Y are two independent discrete random variables, and Z = X + Y, then the probability density function of Z is the discrete convolution of the probability density function of X and the probability density function of Y, which satisfies:

[0220]

[0221] In general, for independent random variables X1,X1,...,X m, then the probability density function of the sum of all random variables is equal to X1,X1,...,X m Discrete convolution of individual probability density functions.

[0222] According to the above discussion, the probability density function H of successful transmission of a data packet in state (i, k) i,k for:

[0223] H i,k =(h0*h1*···*h i )[k] (18)

[0224] Among them, (h0*h1*···*h i )[k] represents h0[k],h1[k],...,h i Discrete convolution of [k].

[0225] For formula (18), we only need to calculate h0[k],h1[k],...,h respectively through FFT algorithm. i [k], then multiply the result and perform an inverse transform to get the final result.

[0226] Next, we find the probability W of a successful data packet transmission i-s,k , according to the conditional probability, we can get

[0227]

[0228] It can be expressed as:

[0229]

[0230] Among them, P l-s is the collision probability of real-time service nodes; S l is the maximum backoff order, i is the backoff order, and k is the number of time slots under the current backoff order.

[0231] Therefore, the probability Q that a real-time service node data packet is successfully transmitted in the i-th order, k-th backoff slot is i [k] can be expressed as W i,k and H i,k The product of, that is:

[0232] Q i [k] = W i-s,k ·H i,k =W i-s,k ·(h0*h1*···*h i )[k] (20)

[0233] (2) Analysis of conditional probability Φ(D|(i,k))

[0234] For the conditional probability Φ(D|(i,k)), we first find the probability distribution of the time it takes for a real-time service node to successfully transmit a data packet in state (i,k), and then integrate the probability density function over (-∞,D).

[0235] make is the MAC delay of a real-time service node successfully transmitting a data packet in the i-th order and k-th time slot. According to the central limit theorem, when the number of time slots k experienced by the node is large enough, It obeys Gaussian distribution, that is

[0236]

[0237] Among them, E i,k for Expectations, D i,k for The variance of .

[0238] Therefore, under the condition that the real-time service node successfully transmits in the i-th order and k-th backoff slot, the probability Φ(D|(i,k)) that the delay does not exceed D can be expressed as:

[0239]

[0240] According to the above formula, we can see that as long as the random variable is determined Expectations i,k and variance D i,k , we can get the probability Φ(D|(i,k)) that the delay of a real-time service node does not exceed D under the condition that the real-time service node successfully transmits in the i-th order and k-th backoff slot. Next, we calculate the expected E i,k and variance D i,k To solve.

[0241] Expectation E i,k Contains the mean of k backoff slots, the duration of i collisions and the duration of data packet transmission, and the variance D i,k It can be expressed as the sum of the variances of k backoff slots, that is:

[0242]

[0243] Among them, E l and D l are the mean and variance of the time slots when the real-time service nodes do not participate in the transmission; T c is the average time slot of collision between nodes when real-time service nodes participate in transmission; T sl is the average length of a data packet transmitted by a real-time service node. Next, we only need to calculate E in the above formulal , D l and T c Three parameters are enough.

[0244] For a single time slot, there are three states in which the time slot is located: idle state, node transmission in the time slot, and node conflict in the time slot. Therefore, the mean value E of a single time slot is l and variance D l Can be expressed as:

[0245]

[0246] Among them, P idle is the channel idle probability, P w is the probability of a WiFi node successfully transmitting, P l-s is the probability of successful transmission of non-real-time service nodes, P c is the probability of conflict between nodes. It should be noted that the above parameters are all real-time service nodes that are in the backoff process and do not participate in the channel competition. According to the state transition model, we can get the expressions as follows:

[0247]

[0248] τ in formula (25) w , n w are the transmission probability and number of WiFi nodes respectively; τ l-o is the transmission probability of the common service node; δ is the length of the idle time slot; T w is the data packet length of the WiFi site; T l-s is the length of the data packet transmitted by real-time LTE-U; T′ c is the average duration of the time slot when the other stations collide when the real-time LTE-U node is in the backoff state, T′ c There are two situations involved.

[0249] (1) At least one WiFi node and a common LTE-U node transmit simultaneously in the same time slot and collide. The probability of this situation can be expressed as:

[0250]

[0251] The corresponding collision duration is max{T w ,T l-o}, where T l-o It is the data packet length for ordinary services.

[0252] (2) Two or more WiFi nodes transmit simultaneously in the same time slot and a collision occurs, and the LTE-U node does not attempt to transmit. The probability of this situation is expressed as:

[0253]

[0254] The corresponding collision duration is T w .

[0255] Combining the above two situations, when the real-time service node is in the backoff state, the average duration of the time slot when the other stations collide is T′ c It can be expressed as:

[0256]

[0257] Finally, we calculate the average collision time T when real-time service nodes participate in competition c When the real-time service nodes in formula (10) participate in the transmission, the average time T of the collision c and T′ c The analysis is similar to that of , but when real-time service nodes participate in the channel competition, there is also a third situation, that is, the probability that ordinary service nodes and real-time service nodes collide at the same time and no WiFi node attempts to transmit. The probability of this situation can be expressed as:

[0258]

[0259] The collision duration corresponding to this situation is T l .

[0260] Therefore, when real-time service nodes participate in transmission, the average time for collision to occur is T c It can be expressed as:

[0261]

[0262] In this way, by substituting equations (25) and (28) into equation (24), the mean value E of the length of a single backoff slot can be calculated: l and variance D l , and then substitute it into formula (23), we can get the expected E of the real-time service node in state (i, k) i,k and variance D i,k Finally, E i,k and D i,k Substituting into formula (22), we can get the probability Φ(D|(i,k)) that the delay of the real-time service node does not exceed D under the condition that the state (i,k) is successfully transmitted. Thus, according to formula (14), the state selection probability Q i The product of [k] and the conditional probability Φ(D|(i,k)) is the probability P(D; i,k) that the real-time service node successfully transmits in state (i,k) with a delay not exceeding D, which can be expressed as:

[0263]

[0264] Among them, F[·] -1 represents the inverse Fourier transform. μ is the delay sensitivity coefficient, S l is the maximum backoff order, W l is the initial backoff window, P l-s is the collision probability of real-time business nodes, is a random variable, E i,k and D i,k They are random variables The expectation and variance of P l-s , E i,k and D i,k The transmission probability τ of the real-time service node l-s is related to l-s Depends on the delay sensitivity coefficient μ, which can be expressed as:

[0265]

[0266] refer to Figure 8 It is concluded that the real-time service node successfully transmits at the state point (i, k), and the MAC delay is no greater than D th The probability P(D th ; i, k); Considering that the successful transmission of a data packet may occur at any state point in the Markov state transition diagram, the delay of successful transmission of a real-time LTE-U node is no more than D th The probability P all It can be expressed as P(D th ; the sum of i, k), that is:

[0267]

[0268] Among them, P(D th ; i, k) is the real-time LTE-U node successfully transmitting in state (i, k) with a delay no greater than D th The probability of backoff; i represents the backoff order, k represents the number of time slots experienced under the current backoff order, S l is the maximum backoff order, W l - s It is the contention window of the real-time LTE-U node.

[0269] For a given delay threshold D th , we need to ensure that the real-time LTE-U node delay calculated by the above formula is no greater than D th The probability P all , which is less than the given constraint probability P outOtherwise, the system cannot guarantee the latency requirements of real-time services using the WiFi band, and users’ speed requirements for cellular networks cannot be met, that is, the following must be met:

[0270]

[0271] Among them, D th is the delay threshold of the real-time business system, indicating the maximum delay that this type of business can accept; P out is the given constraint probability; P(D th ; i, k) is the real-time service node successfully transmitted in state (i, k), and the delay is no more than D th Therefore, real-time services need to satisfy the inequality:

[0272]

[0273] Therefore, for a given constraint probability P out According to the above formula, we get the delay sensitivity coefficient μ, and dynamically adjust the contention window of real-time service nodes to achieve the purpose of reducing the delay of the unlicensed frequency band LTE-U system.

[0274] The present invention calculates the transmission probability τ of WiFi respectively for the number of LTE-U and WiFi in a given system. w and collision probability P w , the transmission probability τ of real-time service nodes sen and collision probability P sen , the transmission probability of ordinary service nodes τ gen and collision probability P gen ; Secondly, according to the given delay threshold D th , calculate the MAC delay of real-time LTE-U is not greater than the threshold D th The probability P all ;Finally, according to the given constraint probability P out , calculate the contention window size generated after the real-time service node conflicts, and thus obtain the value range of the delay sensitivity coefficient μ.

[0275] The embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0276] The embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0277] Therefore, the present invention divides the service types within LTE-U, introduces a delay influencing factor into the delay of real-time service nodes, and introduces a contention window generated by a backoff algorithm, so as to adjust the upper and lower limits of the contention window in real time, and optimize the delay accuracy of real-time services in the LTE system; secondly, in the analysis of the MAC delay of the LTE-U system, the duration of the idle time slot and the busy time slot of the channel are distinguished, thereby improving the accuracy of the delay evaluation of the LTE-U system; finally, by dynamically adjusting the size of the real-time service contention window, the delay requirements of the real-time service are guaranteed, thereby effectively improving the MAC delay of the LTE-U system.

[0278] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0279] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0280] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0281] The above is only a preferred embodiment of the present invention. Without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for optimizing the latency of an unlicensed band LTE system, characterized in that: include: Based on a pre-established system model of a coexistence scenario of an LTE-U system and a WiFi system, the LTE-U nodes and the WiFi nodes in the coexistence scenario system are analyzed through a two-dimensional Markov state transition diagram to obtain state changes of the LTE-U nodes and the WiFi nodes in the system; wherein the system model of the coexistence scenario of the LTE-U system and the WiFi system is established according to a mechanism of LTE accessing a WiFi frequency band; According to the status changes of LTE-U nodes and WiFi nodes in the system, the delay of real-time service under LTE-U node conflict is obtained; According to the delay under the conflict of real-time service LTE-U nodes and the predetermined delay threshold, the probability that all real-time service LTE-U nodes do not exceed the delay threshold is calculated, and the probability that all real-time service LTE-U nodes do not exceed the delay threshold is summed up to obtain the total probability value of all real-time service LTE-U nodes in the system; The total probability value of all real-time service LTE-U nodes is compared with the given constraint probability, and the value range of the delay sensitivity coefficient is obtained according to the total probability value less than the constraint probability; The upper limit and lower limit of the contention window generated by the backoff algorithm are adjusted in real time according to the value range of the delay sensitivity coefficient, so as to optimize the delay accuracy of the real-time service in the LTE system.

2. The method for optimizing the latency of an unlicensed band LTE system according to claim 1, characterized in that: LTE-U system services also include common services.

3. The method for optimizing the latency of an unlicensed band LTE system according to claim 2, characterized in that: The delay of real-time service LTE-U node conflict includes: According to the state changes of LTE-U nodes and WiFi nodes in the system, the single node transmission probability and collision probability of LTE-U nodes and WiFi nodes are obtained respectively; From the transmission probability of a single LTE-U node and the collision probability of a single node, the real-time service LTE-U node and the common service LTE-U node are determined, and the real-time service LTE-U node is calculated to obtain the delay under the conflict of the real-time service LTE-U node; Among them, the state change transfer equation is expressed as: Where i represents the backoff order, k represents the backoff counter value at the current order, P(i,k) represents the probability when the state is (i,k) on the two-dimensional Markov chain, W0 represents the initial backoff window length, and W i represents the backoff window length when the backoff order is i, W w represents the value of the initial backoff window, P w Represents the probability of node collision.

4. The method for optimizing the latency of an unlicensed band LTE system according to claim 3, characterized in that: The calculation process of the transmission probability of a single node is: Substitute the limit steady-state distribution probability formula of the Markov chain into the state change transfer equation to obtain the steady-state probability when the backoff value is 0; Combined with the regularity of Markov chain, the steady-state probability of LTE-U node in any state during the backoff process is obtained; The steady-state probability in any state during the backoff process is classified and discussed by backoff order i, and a simplified formula for the steady-state probability of LTE-U nodes in any state is obtained. When the sum of the steady-state distribution probabilities is 1, the transmission probability of a single node can be obtained by substituting the simplified formula of the steady-state probability under any state of the LTE-U node into the formula; Among them, the limit steady-state distribution probability is defined as: Where b i,k represents the limit steady-state distribution probability, s(t) represents the backoff order of the node at time t, and b(t) represents the value of the backoff counter of the node at time t; The calculation formula of the steady-state probability of the LTE-U node during the backoff process is: Where b i,0 represents the steady-state probability when the state is (i,0), b m,0 represents the steady-state probability when the state is (m,0), b 0,0 represents the steady-state probability at the initial state (i, 0), and M represents the lower limit of the backoff order i; Combined with the regularity of Markov chain, when k∈[0,W i -1], the calculation formula of the steady-state probability of the LTE-U node in any state is: Where, formula (4) to formula (5) b j,0 It represents the steady-state probability when the backoff order is j and the backoff value is 0; When the backoff order is 0<i<M: When the backoff order i = M: Combining formulas (5)-(7), formula (4) is simplified to: When the sum of the steady-state distribution probabilities is 1, the simplified formula for the steady-state probability of the LTE-U node in any state is expressed as: Formula (9) is solved as follows: The transmission probability of a single node is: In the formula, τ w represents the transmission probability of a single node, represents the collision probability at backoff order i, The collision probability when the backoff order is maximum.

5. The method for optimizing the latency of an unlicensed band LTE system according to claim 4, characterized in that: The collision probability of a single node is calculated by calculating the transmission probability τ of a single node w It is inferred that the inference calculation formula is as follows: P w =1-(1-τ w ) n (12) Where n is the number of nodes participating in the competition in the system.

6. The method for optimizing the latency of an unlicensed band LTE system according to claim 5, characterized in that: The calculation method of the probability that all real-time service LTE-U nodes do not exceed the delay threshold includes: According to the state changes of LTE-U nodes and WiFi nodes in the system, the state selection probability and conditional probability of a single node are calculated respectively; The state selection probability and conditional probability of a single node are multiplied to obtain the success of a single real-time service LTE-U node transmission in state (i, k) and the delay does not exceed the delay threshold D th probability; The real-time service LTE-U node is in state (i, k) and the transmission is successful and the delay does not exceed the delay threshold D th The probability calculation formula is: P(D th ;i,k)=Q i [k]·Φ(D th |(i,k)) (13) In the formula, P(D th ; i, k) indicates that the real-time service LTE-U node is successfully transmitted in state (i, k) and the delay does not exceed the delay threshold D th The probability of Q i [k] represents the state selection probability of the real-time service node data packet being successfully transmitted in the i-th order and k-th backoff time slot, Φ(D th |(i,k)) indicates that the real-time service node successfully transmits at the i-th order and k-th backoff slot, and the delay does not exceed D th The conditional probability of .

7. The method for optimizing the latency of an unlicensed band LTE system according to claim 6, characterized in that: The calculation process of state selection probability is as follows: According to the status changes of LTE-U nodes and WiFi nodes in the system, real-time service nodes are screened, and the transmission probability of real-time service nodes is calculated to obtain the collision probability of a single node; Substituting the collision probability of a single node into the conditional probability calculation, we can get the probability of successful transmission of a single data packet; The probability of successful transmission of a single data packet is multiplied by the probability density function of successful transmission of the data packet in state (i, k) to obtain the state selection probability of a single node; The calculation formula for state selection probability is: Q i [k]=W i-s,k ·H i,k (14) In the formula, Q i [k] represents the probability of successful transmission of the real-time service node data packet in the i-th order and k-th backoff time slot, W i-s,k represents the probability of successful transmission of a single data packet, H i,k The probability density function of the successful transmission of a data packet in state (i, k); The formula for calculating the probability of successful transmission of a single data packet is: Where P l-s is the collision probability of real-time service nodes; S l is the maximum backoff order, i is the backoff order, and k is the number of time slots under the current backoff order; The probability density function expression of the successful transmission of a data packet in state (i, k) is: H i,k =(h0*h1*···*h i )[k] (16) In the formula, (h0*h1*···*h i )[k] represents h0[k],h1[k],...,h i Discrete convolution of [k].

8. The method for optimizing the latency of an unlicensed band LTE system according to claim 6, characterized in that: The conditional probability is calculated as follows: Combined with the two-dimensional Markov state transition diagram, the probability of channel idleness, the probability of successful transmission of a single WiFi node, the probability of successful transmission of a non-real-time service node, the probability of conflict between nodes, the data packet length of the WiFi site, the data packet length transmitted by the real-time LTE-U node, and the average duration of the time slot when the other sites collided when the real-time LTE-U node was in the backoff state, the correlation formula was derived; According to the state changes of LTE-U nodes and WiFi nodes in the system, the average duration of the time slot when the real-time service node is in the backoff state and the average time slot time when the real-time service node participates in the transmission are obtained; and the average duration of the time slot when the real-time service node is in the backoff state and the collision between the other stations are substituted into the correlation formula to obtain the mean value E of the length of a single backoff time slot when the real-time service node does not participate in the transmission. l and the variance D of real-time service nodes not participating in transmission l ; By calculating the mean value E of a single backoff slot length l , the variance D of real-time service nodes not participating in transmission l When the real-time service nodes participate in the transmission, the average time slot time of the collision between nodes is substituted into the solution formula of the expectation and variance of a single node to obtain the expectation E of the real-time service node in state (i, k) i,k and variance D i,k ; The expected value E of a real-time service node in state (i, k) i,k and variance D i,k Combined with the central limit theorem, the delay is determined And delay and the expectation E of the corresponding state (i, k) i,k and variance D i,k Substituting into the calculation formula of conditional probability, we can get the condition that the real-time service node successfully transmits in the i-th order and k-th backoff time slot, and the delay does not exceed D th The conditional probability of The derived correlation formulas among the channel idle probability, the probability of successful transmission of a single WiFi node, the probability of successful transmission of a non-real-time service node, and the probability of conflict between nodes are as follows: P c =1-P idle -P w -P l-s (17) In the formula, τ w represents the transmission probability of the WiFi node, n w represents the number of WiFi nodes; τ l-o represents the transmission probability of a common service node, P idle represents the channel idle probability, P w represents the probability of a WiFi node successfully transmitting, P l-s represents the probability of successful transmission of non-real-time service nodes, P c represents the probability of conflict between nodes; When the real-time service node is in the backoff state, the average duration of the time slot when the other stations collide is T′ c The expression is: The average length of a single backoff slot when the real-time service node does not participate in the transmission E l and the variance D of real-time service nodes not participating in transmission l The calculation formula is: Substituting formulas (17) and (18) into (19), we can calculate the mean value E of a single backoff slot length: l and variance D l ;P idle represents the channel idle probability, P w represents the probability of a WiFi node successfully transmitting, P l-s represents the probability of successful transmission of non-real-time service nodes, P c represents the probability of conflict between nodes, T w Indicates the data packet length of the WiFi site; T l-s Indicates the length of the data packet transmitted by real-time LTE-U; T′ c represents the average duration of the time slot when the other sites collide when the real-time service LTE-U node is in the backoff state, δ represents the length of the idle time slot, and E l and D l They represent the mean of the time slots in which the real-time service node does not participate in the transmission and the variance of the time slots in which the real-time service node does not participate in the transmission respectively; When real-time service nodes participate in transmission, the average time slot time T of collision between nodes is c The solution formula is: The expected value E of a real-time service node in state (i, k) i,k and variance D i,k The solution formula is: In the formula, the mean value E of the length of a single backoff slot in formula (19) is l and variance D l and T in formula (20) c Substituting into formula (21), we can get the expected value E of the real-time service node in state (i, k): i,k and variance D i,k , T c It represents the average time slot of collision between nodes when real-time service nodes participate in transmission; T sl Indicates the average length of a data packet transmitted by a real-time service node; According to the central limit theorem, the delay is determined The expression of Gaussian distribution is: The formula for calculating conditional probability is: In the formula, Φ(D th |(i,k)) indicates that the real-time service node successfully transmits at the i-th order and k-th backoff slot, and the delay does not exceed D th The probability of represents the MAC delay of a real-time service node successfully transmitting a data packet at the i-th order and k-th time slot, E i,k for Expectations, D i,k for The variance of .

9. The method for optimizing the latency of an unlicensed band LTE system according to claim 1, characterized in that: Methods for obtaining the value range of the delay sensitivity coefficient according to the total probability value less than the constraint probability include: The expectation E of the real-time business node in state (i, k) i,k and variance D i,k Substituting the conditional probability, we can conclude that when the real-time service node successfully transmits state (i, k), the delay does not exceed D th The probability Φ(D th |(i,k)) value, and then deduce that the real-time service node is in state (i,k) and the transmission is successful and the delay does not exceed D th The probability P(D th ; i, k); The delay of successful transmission of real-time service LTE-U node in all states is no more than D th The probability P(D th ; the sum of i,k)P all Compare with the given constraint probability and select P(D th ; i, k); For all the screened P(D th ; i, k) The delay sensitivity coefficient μ is solved to obtain the value range of the delay sensitivity coefficient; Among them, for all the screened P(D th ; i, k) are used to derive the formula: In the formula, F[·] -1 represents the inverse Fourier transform; μ represents the delay sensitivity coefficient, S l represents the maximum backoff order, W l represents the initial backoff window, P l-s represents the collision probability of real-time business nodes, represents a random variable, E i,k and D i,k Denote random variables The expectation and variance of P l-s , E i,k and D i,k The transmission probability τ of the real-time service node l-s Regarding l-s Depends on the delay sensitivity coefficient μ, and then derives the delay sensitivity coefficient μ, and the expression of the delay sensitivity coefficient μ is as follows: The delay of successful transmission of real-time service LTE-U node in all states is no more than D th The probability P(D th ; the sum of i,k)P all The expression is: In the formula, P(D th ; i, k) is the real-time LTE-U node successfully transmitting in state (i, k) with a delay no greater than D th The probability of i represents the backoff order, k represents the number of time slots experienced under the current backoff order, S l is the lower limit of the backoff order, W l-s is the contention window of the real-time LTE-U node; Screen out P(D th ; The expression of i,k) is: Where P out represents the given constraint probability, represents the collision probability of real-time LTE-U nodes under any backoff order, It represents the collision probability of real-time LTE-U node under the maximum backoff order.

10. The method for optimizing the latency of an unlicensed band LTE system according to claim 1, characterized in that: The method of adjusting the upper limit and lower limit of the contention window generated by the backoff algorithm in real time includes: The delay sensitivity coefficient is introduced into the binary exponential backoff algorithm, so that the binary exponential backoff algorithm can adjust the contention window based on the delay sensitivity; The formula for adjusting the contention window is: Where μ is the delay sensitivity coefficient, m l is the maximum backoff order, W l is the size of the initial contention window.

Images