High energy efficiency transmission method in a cell-free massive urllc system

Abstract

The application discloses a high energy efficiency transmission method in a cell-free massive URLLC system, which comprises the following steps: (1) an AP performs minimum mean square error channel estimation according to an uplink pilot signal to obtain an estimated channel state information vector, and returns the vector to a CPU; (2) according to the obtained estimated channel state information, the CPU controls an AP downlink transmission power strategy and a user association strategy with the maximum system energy efficiency as a target; (3) the AP performs downlink data transmission under the control of the CPU, and calculates a signal-to-interference-and-noise ratio and a reachable rate of a user end; (4) the user feeds back the reachable rate to the AP, the CPU further constructs and solves an energy efficiency optimization problem to jointly optimize the power control strategy and the user association strategy, and feeds back the updated strategy to the AP. The application meets the user communication reliability and real-time requirement, and improves the system energy efficiency.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a high energy efficiency transmission method in a cell-free massive URLLC system. BACKGROUND

[0002] URLLC is an important technology in the fifth generation mobile communication system, and its main goal is to achieve low latency, high energy efficiency and high reliability communication. In the process of realizing URLLC, there is a mutual restraining relationship among latency, rate and reliability, which needs to be balanced and optimized among various technical challenges. In terms of energy efficiency, energy saving is a global challenge, and high energy efficiency communication can reduce energy consumption and operating costs. However, there is a mutual restraining relationship between latency and energy efficiency, because reducing latency usually requires more energy consumption, and improving energy efficiency may increase communication latency.

[0003] Cell-free massive MIMO is one of the important ways to realize URLLC. Compared with traditional centralized massive MIMO systems, cell-free massive MIMO systems use a large number of small antennas distributed in the environment for data transmission and reception, which realizes better spatial diversity and spectrum efficiency. This architecture can reduce communication latency, improve signal quality and reliability, and has the potential to significantly improve energy efficiency. SUMMARY

[0004] The application aims at the problems existing in the prior art, and proposes a high energy efficiency transmission method in a cell-free massive URLLC system, which comprises the following steps:

[0005] (1) The AP performs minimum mean square error channel estimation according to the uplink pilot signal to obtain an estimated channel state information vector, and returns it to the CPU;

[0006] (2) According to the estimated channel state information obtained, the CPU controls the AP downlink transmission power strategy and user association strategy with the goal of maximizing system energy efficiency;

[0007] (3) The AP performs downlink data transmission under the control of the CPU, and calculates the signal-to-interference-and-noise ratio and achievable rate of the user end;

[0008] (4) The user feeds back the achievable rate to the AP, and the CPU further constructs and solves the energy efficiency optimization problem to jointly optimize the power control strategy and user association strategy, and feeds back the updated strategy to the AP.

[0009] Further, step (1) is as follows:

[0010] (1-1) Users send their pilot signals to AP synchronously, and the uplink channel is estimated by AP using minimum mean square error estimation method. If the pilots of different users are orthogonal to each other, the variance of the estimated channel vector element is:

[0011]

[0012] where, ν ik represents the variance of the estimated channel vector element, represents the estimated channel vector between the ith AP and the kth user, and the subscript n represents the nth element in the vector, l ik represents the large-scale fading factor between the ith AP and the kth user, ψ k represents the pilot sequence of the kth user, τ ul represents the pilot length, ρ ul represents the uplink transmit signal-to-noise ratio;

[0013] (1-2) AP uploads the locally estimated channel state information to CPU through the backhaul link, and then CPU obtains the global channel state information of the system.

[0014] Further, step (2) is specifically as follows:

[0015] (2-1) CPU sends the solution results of the energy efficiency optimization problem of the last round, i.e. the transmit power of each AP and its associated user, to the corresponding AP through the front transmission link. After receiving this information, the AP makes the corresponding update. If this round is the first round of optimization, CPU uses the equal power allocation strategy, and all APs use the user association strategy of serving users simultaneously.

[0016] Further, the step (3) specifically includes:

[0017] (3-1) AP uses maximum ratio transmission precoding strategy to send downlink signal to user, which can be represented as:

[0018]

[0019] where represents the received signal of the kth user, p dl represents the maximum transmit power of the AP, N represents the total number of APs in the system, represents the user set associated with the ith AP, h ik represents the channel between the ith AP and the kth user, ι ik represents the power control factor between the ith AP and the kth user, represents the estimated channel, x k represents the data symbol sent to the kth user, represents white noise satisfying complex Gaussian distribution;

[0020] (3-2) The kth user calculates the achievable rate according to the received signal, which can be expressed as:

[0021]

[0022] wherein R k represents the achievable rate of the kth user, γ k represents the signal-to-interference-plus-noise ratio of the kth user, V represents channel dispersion, n represents the transmission block length, ∈ represents the average decoding error probability, and Q(·) represents the tail distribution function of the standard normal distribution.

[0023] Further, the step (4) specifically comprises:

[0024] (4-1) The CPU constructs the energy efficiency according to the received achievable rate:

[0025]

[0026] wherein B represents the transmission bandwidth, β represents the proportion of downlink data transmission time, α represents the drain efficiency of the power amplifier, M represents the number of antennas equipped by each AP, P tc represents the power consumption of the transceiver radio frequency chain, P CDBT represents the power consumption proportional to the rate, In addition, wherein ψ k represents the pilot sequence of the kth user, ν ik represents the variance of the estimated channel between the ith AP and the kth user, l ik represents the large-scale fading coefficient between the ith AP and the kth user;

[0027] (4-2) The CPU models the energy efficiency optimization problem as:

[0028]

[0029]

[0030]

[0031]

[0032] wherein c ik ∈{0,1} represents whether the ith AP is associated with the kth user, represents the minimum rate requirement of the kth user. The CPU equivalently converts this problem into a second-order cone programming problem, solves it through the optimization toolbox of MATLAB, and sends the updated power control coefficient and association index to the AP.

[0033] The beneficial effects of the present application are:

[0034] The present application effectively solves the problem of mismatch between the traditional Shannon capacity and the rate representation in URLLC short packet communication by using the limited block length coding theory. Meanwhile, compared with the centralized architecture, the performance of the spectral efficiency and energy efficiency is improved by using the user-centered service feature of the non-cell large-scale MIMO architecture. The present application solves the energy efficiency optimization problem by jointly optimizing the AP downlink power and user association strategy, and takes into account the delay, reliability and rate requirements of different users. Compared with the existing scheme which assumes that users are served by a fixed number or proportion of APs, the present application further improves the flexibility of service and reduces the multi-user interference, and has good application potential in the delay-sensitive business scenario with limited energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a flowchart of a high energy efficiency transmission method in a non-cell large-scale URLLC system according to the present application;

[0036] Figure 2 is a simulation result graph of the energy efficiency of a high energy efficiency transmission method in a non-cell large-scale URLLC system according to the present application. DETAILED DESCRIPTION

[0037] In order to deepen the understanding and understanding of the present application, the following will introduce the scheme in detail in combination with the drawings.

[0038] Embodiment 1: The present embodiment provides a high energy efficiency transmission method in a non-cell large-scale URLLC system, as shown in Figure 1 , which comprises:

[0039] Assuming that there are N multi-antenna APs and K URLLC users in a non-cell large-scale MIMO system, each AP is equipped with M antennas, and they provide services for users on the same time-frequency resource.

[0040] (1) The AP performs minimum mean square error channel estimation according to the uplink pilot signal to obtain the estimated channel state information vector, and returns it to the CPU.

[0041] This step specifically includes:

[0042] (1-1) The users synchronously send their pilot signals to the AP, and the uplink channel is estimated by the AP using the minimum mean square error estimation method. Assuming that the pilot signals of different users are orthogonal to each other, the variance of the estimated channel vector elements is represented as:

[0043]

[0044] wherein, v ik denotes the variance of the estimated channel vector element, denotes the estimated channel vector between the ith AP and the kth user, and the subscript n denotes the nth element within the vector, l ik denotes the large-scale fading factor between the ith AP and the kth user, ψ k denotes the pilot sequence of the kth user, τ ul denotes the pilot length, ρ ul denotes the uplink transmit signal-to-noise ratio;

[0045] (1-2) The AP uploads the locally estimated channel state information to the CPU through a backhaul link, and then the CPU obtains the system global channel state information.

[0046] (2) According to the estimated channel state information obtained by aggregation, the CPU controls the AP downlink transmission power strategy and user association strategy with the goal of maximizing system energy efficiency;

[0047] This step specifically includes:

[0048] (2-1) The CPU sends the solution results of the energy efficiency optimization problem of the last round, i.e., the transmission power of each AP and its associated users, to the corresponding AP through the front link. After receiving this information, the AP makes the corresponding update. If this round is the first round of optimization, the CPU adopts the equal power allocation strategy, and all APs adopt the user association strategy of serving users simultaneously.

[0049] (3) The AP performs downlink data transmission under the control of the CPU and calculates the signal-to-interference-and-noise ratio and achievable rate at the user end.

[0050] This step specifically includes:

[0051] (3-1) The AP uses the maximum ratio transmission precoding strategy to send downlink signals to users, which can be represented as:

[0052]

[0053] wherein denotes the received signal of the kth user, p dl denotes the maximum transmission power of the AP, denotes the set of users associated with the ith AP, h ik denotes the channel between the ith AP and the kth user, ι ik the power control factor between the ith AP and the kth user, denotes the estimated channel, x k denotes the data symbol sent to the kth user, denotes white noise satisfying complex Gaussian distribution;

[0054] (3-2) The kth user acquires statistical channel information by using channel sensing technology, and calculates the signal-to-interference-and-noise ratio γ using channel hardening characteristics k , which is expressed as:

[0055]

[0056] wherein, η k =[η 1k ,...,η Mk ] T denotes a power control factor vector and C k =diag(c 1k ,...,c Nk ) denotes the kth user association matrix. Specifically, c ik =1 indicates that the ith AP and the kth user establish association, and c ik =0 indicates disassociation;

[0057] (3-3) The kth user calculates the achievable rate according to its signal-to-interference-and-noise ratio, which can be expressed as:

[0058]

[0059] wherein R k denotes the achievable rate of the kth user, γ k denotes the signal-to-interference-and-noise ratio of the kth user, V denotes channel dispersion, n denotes the transmission block length, ∈ denotes the average decoding error probability, and Q(·) denotes the tail distribution function of the standard normal distribution.

[0060] (4) The user feeds back the achievable rate to the AP, and the CPU further constructs and solves the energy efficiency optimization problem to jointly optimize the power control strategy and the user association strategy, and feeds back the updated strategy to the AP.

[0061] This step (4) specifically includes:

[0062] (4-1) The CPU constructs the energy efficiency according to the received achievable rate:

[0063]

[0064] wherein B denotes the transmission bandwidth, β denotes the proportion of downlink data transmission time, α denotes the power amplifier drain efficiency, M denotes the number of antennas equipped by each AP, P tc denotes the power consumption of the transceiver radio frequency chain, and P CDBT denotes the power consumption proportional to the rate;

[0065] (4-2) The CPU models the energy efficiency optimization problem as:

[0066] P1:

[0067] s.t.C1.1:

[0068] C1.2:

[0069] C1.3:

[0070] where R k denotes the achievable rate of the kth user, denotes the minimum rate of the kth user;

[0071] (4-3) The CPU converts this problem equivalently into a second-order cone programming problem. To eliminate the multiplicative coupling between η ik and c ik , the following constraints are introduced to the original optimization problem:

[0072] C1.4:

[0073] C1.5:

[0074] C1.6:

[0075] where q ik is an auxiliary variable. The above constraints can ensure that when c ik = 0, there is η ik = 0, i.e., when the ith AP and the kth user do not establish association, the power control coefficient between them also implicitly approaches 0;

[0076] (4-4) The CPU further uses the perspective method to equivalently convert the original optimization problem P1 into:

[0077] P2:

[0078] s.t.C2.1:

[0079] C2.2:

[0080] C2.3:

[0081] C1.2-C1.6, where t k and t0 are optimization variables introduced by the perspective method, is a tight lower bound of R k , and is a decreasing convex function with respect to x;

[0082] (4-5) For non-convex rate constraint C2.1, CPU exploits the monotonicity of f(x) and introduces an auxiliary variable u k which is transformed into

[0083] C2.1.1: γ k ≥ u k ,

[0084] C2.1.2: where constraint C2.1.1 is equivalent to Since the left side of the inequality is a quadratic function divided by a linear function, for η k and u k are jointly convex. Further, by first-order Taylor expansion of the left side of the inequality, the constraint is further transformed into

[0085] C2.1.1.1:

[0086] where,

[0087]

[0088] where η n and denote η and u k after n consecutive convex approximation iterations.

[0089] Similarly, for non-convex constraint C2.1.2, its convex lower bound is obtained by using the successive convex approximation, denoted as

[0090]

[0091] where, Thus, constraint C2.1.2 can be further transformed into

[0092] C2.1.2.1: When at the n+1th optimization iteration, problem P2 is equivalently transformed into

[0093] P3:

[0094] s.t. C1.2-C1.6, C2.2, C2.3, C2.1.1.1, C2.1.2.1,

[0095] All constraints in problem P3 are linear constraints or second-order cone constraints, and the optimization variables contain both continuous and binary integer variables, thus P3 is a mixed integer second-order cone programming problem;

[0096] (4-6) Although P3 can be solved directly, its solution complexity typically increases exponentially with the number of access points (APs) and users in the system. To overcome this problem, a continuous relaxation technique is used to reduce the complexity of P3 by c. ik Transform into And transform constraint C1.4 into:

[0097]

[0098] in, It is after rounds of iteration Replacing constraint C1.4 in P3 with constraint C1.4.1, while keeping other constraints unchanged, problem P3 becomes a second-order cone programming problem, which can be solved efficiently using MATLAB's optimization toolbox. Subsequently, the solved... Using discontinuous mappings: if Then map it to 1, otherwise map it to 0;

[0099] (4-7) Send the updated power control coefficients and correlation coefficients to the AP.

[0100] Appendix Figure 2 In the simulation results, BnB (Branch and Bound) is a branch and bound method used to find the performance upper bound of the original problem. Canonical (published in IEEE Transactions on Wireless Communications in 2017 by Hien Quoc Ngo et al., titled "Cell-Free Massive MIMO Versus Small Cells") is a traditional cellless architecture employing equal power allocation and a fully associative strategy. Algorithm 1 and Algorithm 2 are the methods of this patent. Algorithm 1 directly solves the mixed-integer second-order cone programming problem in steps (4-5), while Algorithm 2 solves the second-order cone programming problem in steps (4-6). It is evident that the method of this invention achieves higher energy efficiency than Canonical and is also close to the energy efficiency upper bound obtained by BnB.

[0101] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1.A method for high energy efficiency transmission in a cell-free massive URLLC system, the method comprising: The method comprises the following steps: (1) the AP makes minimum mean square error channel estimation according to the uplink pilot signals to obtain an estimated channel state information vector and returns it to the CPU; (2) according to the estimated channel state information obtained, the CPU controls the AP downlink transmission power strategy and user association strategy with the maximum system energy efficiency as the target; (3) the AP performs downlink data transmission under the control of the CPU and calculates the signal-to-noise ratio and achievable rate of the user end; (4) the user feeds back the achievable rate to the AP, and the CPU further constructs and solves the energy efficiency optimization problem to jointly optimize the power control strategy and the user association strategy, and feeds back the updated strategy to the AP, as follows: (4-1) the CPU constructs the energy efficiency according to the received achievable rate: , wherein, denotes a transmission bandwidth, denotes a downlink data transmission time ratio, denotes a power amplifier drain efficiency, denotes a number of antennas equipped by each AP, denotes a transceiver radio frequency chain power consumption, denotes a power consumption proportional to a rate, further, wherein, denotes a pilot sequence of the th user, denotes a variance of an estimated channel between the th AP and the th user, denotes a large-scale fading coefficient between the th AP and the th user; (4-2) the CPU models the energy efficiency optimization problem as: wherein, denotes whether an association is established between the AP and the user, denotes the rate of the user, denotes the minimum rate requirement of the user, for this mixed integer nonlinear programming problem, CPU firstly converts it into a mixed integer second order cone programming problem using continuous convex approximation technique, secondly, in order to reduce the solving complexity, uses continuous relaxation and discontinuous mapping technique to further equivalently convert this mixed integer second order cone programming problem into a second order cone programming problem, solves it through MATLAB optimization toolbox, and sends the updated power control coefficient and association coefficient to the AP. 2.The high energy efficiency transmission method in a cell-free massive URLLC system according to claim 1, wherein: Step (1) is as follows: (1-1) the users synchronously send their pilot signals to the AP, and the uplink channel is estimated by the AP using the minimum mean square error estimation method, and the variance of the estimated channel vector elements is represented as: , wherein, denotes the variance of the estimated channel vector element, denotes the estimated channel vector between the th AP and the th user, and the subscript denotes the th element within the vector, denotes the large scale fading factor between the th AP and the th user, denotes the pilot sequence of the th user, denotes the pilot length, denotes the uplink transmit power; (1-2) the AP uploads the locally estimated channel state information to the CPU through the backhaul link, and then the CPU obtains the global channel state information of the system. 3.The high energy efficiency transmission method in a cell-free massive URLLC system according to claim 1, wherein: Step (2) is as follows: (2-1) the CPU sends the solution results of the energy efficiency optimization problem of the last round, i.e. the transmission power of each AP and its associated users, to the corresponding AP through the front-end link, and the AP makes the corresponding update after receiving the information, if it is the first round of optimization, the CPU adopts the equal power allocation strategy, and all APs adopt the user association strategy that all users are served simultaneously. 4.The high energy efficiency transmission method in a cell-free massive URLLC system of claim 1, wherein: Step (3) is as follows: (3-1) the AP uses the maximum ratio transmission precoding strategy to send downlink signals to the user, which is represented as: ， wherein denotes the received signal of the th user, denotes the maximum transmit power of the AP, denotes the total number of APs in the system, denotes the set of users associated with the th AP, denotes the channel between the th AP and the th user, the power control factor between the th AP and the th user, denotes the estimated channel, denotes the data symbol transmitted to the th user, denotes white noise satisfying complex Gaussian distribution; (3-2) The first user calculates the achievable rate from its received signal, denoted as: , wherein denotes the achievable rate of the denotes the signal-to-interference-and-noise ratio of the denotes the channel dispersion, denotes the transmission block length, denotes the average decoding error probability, denotes the tail distribution function of the standard normal distribution.​​

Images