A resource allocation method for high-speed mobile terminals and low-speed user coexistence networks

Abstract

This invention belongs to the field of wireless communication technology, specifically relating to a resource allocation method for a network where high-speed mobile terminals and low-speed users coexist. The method includes: constructing a NOMA system model based on S-OTFS; clustering and pairing high-speed and low-speed mobile users (H-UEs) and low-speed users (L-UEs); generating sparse patterns and mapping H-UE symbols to the DD domain, then performing ISFFT processing to generate sparse OTFS signals; mapping L-UE symbols to the TF domain and performing SC-FDMA modulation to generate SC-FDMA signals; a NOMA signal composed of the two signals from the base station; both types of users receiving the NOMA signal and calculating the spectral efficiency of each user; constructing an optimization problem to maximize the average spectral efficiency of H-UEs based on the user's spectral efficiency; solving the optimization problem to obtain a resource allocation scheme; and the system transmitting signals according to the resource allocation scheme. This invention significantly improves the communication system, rate, and spectral efficiency in heterogeneous mobility scenarios.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically relating to a resource allocation method for a network where high-speed mobile terminals and low-speed users coexist. Background Technology

[0002] With the development of 5G / 6G technologies, the coexistence of high-speed mobile scenarios (such as high-speed trains and drones) and low-speed fixed scenarios (such as IoT devices and fixed users) is becoming increasingly common. In millimeter-wave communication, due to the short wavelength characteristics of the high-frequency band, narrow beam alignment is required, but frequent beam training increases overhead and is sensitive to high-speed movement. Simultaneously, the severe Doppler effect introduced by high-speed movement can disrupt orthogonality, affecting the reliability of traditional technologies such as OFDM. Orthogonal Time-Frequency-Spacetime (OTFS) modulation converts the time-varying channel into a quasi-static channel in the delay-Doppler (DD) domain, effectively suppressing Doppler frequency shift and making it more reliable in high-mobility environments. However, in scenarios with multiple users sharing spectrum and power constraints, traditional OTFS often results in excessively high power spectral density due to its non-sparse mapping of the entire bandwidth, reducing energy efficiency and exacerbating interference to other users.

[0003] Non-orthogonal multiple access (NOMA) technology enables high-speed and low-speed users to share time-frequency resources through power domain multiplexing, improving access efficiency in heterogeneous mobility environments. Existing research shows that OTFS-based NOMA schemes can help meet the needs of both high-speed and low-speed users simultaneously. However, in millimeter-wave NOMA systems, existing beamwidth designs and power allocation schemes are insufficient in suppressing interference for high-speed users, making it difficult to simultaneously ensure spectral efficiency for high-speed users and reliability for low-speed users. Especially in power-constrained environments, typical OTFS-NOMA schemes do not fully utilize the sparsity of the DD domain, failing to effectively reduce interference and energy consumption.

[0004] In recent years, existing resource studies have generally focused on improving overall spectral efficiency using OTFS-NOMA schemes. However, in order to eliminate inter-symbol interference in the DD domain, relatively complex equalization techniques are employed, making it difficult to achieve fast real-time power allocation for high-mobility channels. Therefore, there is still room for improvement in energy efficiency and interference suppression capabilities in mixed high-speed and low-speed user scenarios. How to perform reasonable resource gridding and power allocation within this cross-domain OTFS-NOMA framework, and fully utilize the channel sparsity of OTFS in the delay-Doppler domain to maximize system throughput and spectral efficiency, is an urgent problem to be solved. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a resource allocation method for a network where high-speed mobile terminals and low-speed users coexist. The method includes:

[0006] S1: Construct a NOMA system model based on S-OTFS;

[0007] S2: Cluster and pair high-speed mobile users (H-UEs) with low-speed mobile users (L-UEs) to obtain multiple NOMA groups; allocate subbands for each NOMA group;

[0008] S3: Generate a sparse pattern and map the H-UE symbols to the DD domain according to the sparse pattern. After ISFFT processing, generate a sparse OTFS signal; map the L-UE symbols to the TF domain and perform SC-FDMA modulation to generate an SC-FDMA signal; the base station transmits a NOMA signal composed of the OTFS signal and the SC-FDMA signal.

[0009] S4: H-UE users and L-UE users receive NOMA signals and calculate the spectral efficiency of H-UE users and L-UE users;

[0010] S5: Construct an optimization problem to maximize the average spectral efficiency of H-UE based on the spectral efficiency of H-UE users and L-UE users;

[0011] S6: Solve the optimization problem of maximizing the average spectral efficiency of H-UE to obtain the resource allocation scheme; the system transmits signals according to the resource allocation scheme.

[0012] Preferably, the NOMA system model based on S-OTFS includes a base station (BS) and... Users are categorized into high-speed mobile users (H-UE) and low-speed mobile users (L-UE) based on their mobile speed. H-UE uses sparse OTFS modulation for its data symbols, while L-UE uses SC-FDMA modulation for its data symbols.

[0013] Preferably, the process of clustering and pairing high-speed mobile user (H-UE) and low-speed mobile user (L-UE) includes: clustering an H-UE group and an L-UE group into a NOMA group, wherein each NOMA group contains H-UE users and L-UE users.

[0014] Preferably, the process of generating a sparse pattern and mapping H-UE symbols to the DD domain based on the sparse pattern includes:

[0015] Determine the sparsity of the current transmission;

[0016] Define a data symbol sequence index based on sparsity; generate a sparse pattern based on the data symbol sequence index;

[0017] Construct a sparse mask matrix based on the sparse pattern;

[0018] The H-UE symbols are mapped to the DD field based on the sparse mask matrix.

[0019] Furthermore, the sparsity is the largest integer satisfying the condition, which is expressed as:

[0020]

[0021]

[0022] Where K represents sparsity. This indicates the interference margin protection factor. This represents the interference channel gain from the base station to the L-UE. This indicates the maximum interference power that the L-UE can tolerate. The value represents the transmit power of the H-UE, N represents the Doppler grid size, and M represents the time delay dimension grid size.

[0023] Furthermore, the sparse pattern representation generated based on the data symbol sequence index is as follows:

[0024]

[0025] in, This represents the data symbol sequence index, N represents the Doppler grid size, and M represents the time delay dimension grid size. Indicates Doppler index, Indicates a delay index. and These represent the Doppler step size factor and the time delay step size factor, respectively. and These are represented as Doppler random offset and time delay random offset, respectively. This indicates taking the modulus.

[0026] Preferably, the formula for calculating the spectral efficiency of H-UE users and L-UE users is as follows:

[0027]

[0028]

[0029] in, Indicates the first Group H-UE Spectral efficiency for individual users. Indicates the first In group L-UE, the first The user in the first Spectral efficiency across frequency bands W represents the number of users in each user group, and W represents the number of TF resource blocks occupied by each L-UE. Indicates the total gain of H-UE. This represents the total gain of L-UE. Indicates that the base station sends to the first Group H-UE The power of the signal transmitted by each user. Indicates the number of L-UE user groups. Indicates the first Group H-UE and the first Pairing variables for L-UE pairings, Indicates the first In group L-UE, the first Individual users Time-frequency plane channel gain on the carrier band, Indicates that the base station sends to the first Group 1 The power of the transmitted signal of each L-UE user, Indicates the first The average transmit power of group L-UE, This represents the noise variance.

[0030] Preferably, the optimization problem for maximizing the average spectral efficiency of H-UE is expressed as:

[0031]

[0032] in, , and These represent the power variable set, the beamwidth set, and the user equipment scheduling variable set, respectively. Indicates the total duration of beam alignment and communication. This indicates the beam alignment time, and W represents the number of TF resource blocks occupied by each L-UE. Indicates the first Group H-UE Spectral efficiency for individual users. Indicates the subcarrier spacing. This indicates the maximum transmit power of the base station. Indicates that the base station sends to the first The first L-UE The power of the signal transmitted by each user. Indicates the first The average transmit power of group L-UE, Indicates that the base station sends to the first Group H-UE The power of the signal transmitted by each user. This represents the minimum achievable beamwidth. This indicates the maximum achievable beamwidth. Indicates the first The NOMA group used the first Scheduling variables for each frequency band Indicates the first Group H-UE and the first Pairing variables for L-UE pairings, Indicates the first In group L-UE, the first The user in the first Spectral efficiency across frequency bands This represents the minimum throughput required for each L-UE. Indicates the number of H-UE user groups. Indicates the number of frequency band indices. Indicates the number of L-UE user groups. This indicates the number of users in each user group. Represents the H-UE user group index set. Represents the L-UE user group index set. Represents a set of frequency band indices. This represents the set of user indexes within each user group.

[0033] Preferably, the process of solving the optimization problem of maximizing the average spectral efficiency of H-UE includes:

[0034] The optimization problem of maximizing the average spectral efficiency of H-UE is decoupled into a scheduling subproblem and a beam-power joint allocation subproblem. Under the condition of fixing some variables, the other part of the variables is optimized. By using variable relaxation, relaxation variable introduction and continuous convex approximation methods, the non-convex constraints and nonlinear objective functions of the two subproblems are transformed into convex approximation problems. The convex approximation problems are solved to obtain the stage-optimal solution.

[0035] Alternating iterative updates are performed between the two sub-problems. In each iteration, a convex approximation model is reconstructed and solved based on the current solution until the objective function value or variable change satisfies the preset convergence condition. Finally, the obtained continuous solution is binarized or quantized to determine the frequency band scheduling results, beamwidth, and transmit power allocation scheme of each NOMA group, i.e., the resource allocation scheme.

[0036] The beneficial effects of this invention are as follows:

[0037] This invention constructs a discretized sparse mapping in the delay-Doppler domain, enabling high-speed user signals to occupy only a portion of the DD grid resources. This reduces the power spectral density per unit bandwidth from the source, significantly weakening co-channel interference to low-speed users and improving spectrum utilization efficiency under conditions of coexistence of high and low-speed users. The original mixed-integer non-convex problem is decomposed into a scheduling subproblem and a beam / power optimization subproblem, and iteratively solved using relaxation variables and continuous convex approximation methods. This ensures numerical convergence and realizability, avoiding the problems of difficulty in convergence or excessive complexity in direct solutions in existing technologies. Considering the joint model of beam alignment overhead and beamwidth influence, this invention, through the collaborative design of sparse transmission and power optimization, improves the effective throughput performance of high-speed users in millimeter-wave directional communication environments while ensuring the quality of service for low-speed users. Attached Figure Description

[0038] Figure 1 This is a flowchart of the resource allocation method for a network where high-speed mobile terminals and low-speed users coexist in this invention.

[0039] Figure 2 This is a schematic diagram of the NOMA system model based on S-OTFS in this invention;

[0040] Figure 3 This is a schematic diagram of the discrete sparse pattern in the DD domain of this invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] This invention proposes a resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, such as... Figure 1 As shown, the method includes the following:

[0043] S1: Construct a NOMA system model based on S-OTFS.

[0044] Construct a NOMA system model based on S-OTFS. This system includes a base station (BS) and a user set. The user set is divided into two mutually exclusive subsets based on user mobility speed: the high-speed user set is denoted as... The set of low-speed users is denoted as Due to high-speed movement, the channel exhibits drastic time-varying characteristics, resulting in significant Doppler frequency shift. This disrupts the subcarrier orthogonality of traditional OFDM systems. Therefore, in this invention, the high-speed user (H-UE) employs sparse orthogonal time-frequency-space (S-OTFS) modulation technology, utilizing its robustness in the DD domain to combat the bi-dispersive channel. The low-speed user aggregated channel is relatively stable, mainly affected by multipath effects and shadowing fading. To reduce terminal complexity and peak-to-average power ratio (PAPR), the low-speed user (L-UE) employs single-carrier frequency division multiple access (SC-FDMA) modulation technology, such as... Figure 2 As shown.

[0045] Due to the high-speed movement of the H-UE, the millimeter-wave channel changes rapidly in time and exhibits frequency selectivity. This dual-dispersion characteristic is most sparse in the DD domain. (Base station and...) individual users Time-varying channel impulse response It can be modeled as Superposition of physical propagation paths:

[0046]

[0047] in, The total number of multipath paths is a factor; millimeter-wave channels exhibit sparsity, typically... Smaller For the first The path complex gain includes large-scale fading and antenna array gain. For the first Transmission delay of each path, For the first The Doppler frequency shift of the path, for a moving speed of users, ,in For the first The angle between the arrival angle of the path and the mode of motion.

[0048] Through the symplectic finite Fourier transform, the physical channel can be mapped to the DD domain, and the DD domain channel response function... Represented as:

[0049]

[0050] For L-UE, The channel primarily exhibits frequency-selective fading, meaning it is mainly distributed along the time delay axis in the DD domain and concentrated near zero frequency on the Doppler axis. For H-UE, Significant and potentially different along different paths, the channel in the DD domain is characterized by being scattered across different paths. The sparse pulses on the coordinates, this dispersion in two dimensions is the manifestation of bichromatic dispersion, and it is also the key to distinguishing low-speed signals from high-speed signals in the DD domain demodulation.

[0051] S2: Cluster and pair high-speed mobile users (H-UEs) with low-speed mobile users (L-UEs) to obtain multiple NOMA groups; allocate subbands to each NOMA group.

[0052] The total system bandwidth is divided into There are orthogonal sub-bands, and the index set is... The index is And assume By adopting the S-OTFS-NOMA scheme, it is necessary to cluster an H-UE group and an L-UE group into a NOMA group, and each NOMA group contains H-Ues and Let there be L-UEs, and let This represents the H-UEs / L-UEs in this NOMA group. Each H-UE and L-UE occupies a space in the DD field. In adjacent resource blocks and in the time-frequency domain There are 1 adjacent resource blocks, of which This means that H-UEs use Orthogonal Multiple Access (OMA) in the DD domain, L-UEs use OMA in the TF domain, and NOMA is applied between H-UEs and L-UEs in the TF domain. The system simultaneously utilizes time-invariant channel gain in the delay-Doppler plane and time-varying channel gain in the time-frequency plane to serve both the H-UE and L-UE groups. In this way, a large amount of time-frequency resource blocks occupied by H-UEs can be shared with L-UEs, thereby improving spectrum efficiency.

[0053] To achieve NOMA transmission, this embodiment defines a specific scheduling variable: user pairing variable. and sub-band assignment variables , Represents a binary variable used to indicate the first... Is group H-UE related to the first Group L-UEs are paired to form a NOMA cluster, that is:

[0054]

[0055] Among them, when At that time, the base station will overlay transmissions in the same time-frequency block, power domain, and beam space. and The signals form NOMA interference groups, and the receiver performs serial interference cancellation based on power conditions. To simplify decoding complexity and ensure NOMA gain, this embodiment specifies that each NOMA cluster strictly includes one group of H-UEs and one group of L-UEs, that is:

[0056]

[0057] Represents a binary variable used to indicate the first... Are group H-UE and its paired NOMA clusters assigned to the first... Sub-band, that is:

[0058]

[0059] This variable determines the frequency domain location of the NOMA cluster, if and This means It was also implicitly scheduled to the sub-band. Above. Furthermore, this embodiment specifies a one-to-one correspondence between each NOMA group and a frequency band, that is:

[0060]

[0061] S3: Generate a sparse pattern and map the H-UE symbols to the DD domain according to the sparse pattern. After ISFFT processing, generate a sparse OTFS signal; map the L-UE symbols to the TF domain and perform SC-FDMA modulation to generate an SC-FDMA signal; the base station transmits a NOMA signal composed of the OTFS signal and the SC-FDMA signal.

[0062] Unlike traditional OTFS, which fills the entire DD domain grid with data symbols, S-OTFS places non-zero data symbols only at specific locations in the DD domain, leaving the rest as zero. This sparse mapping is a prerequisite for reducing interference to L-UEs and achieving efficient resource allocation. In this system, the information symbols for each H-UE group are mapped onto the delay-Doppler DD plane. This plane has the maximum delay. (unit: seconds) and maximum Doppler shift (Unit: Hertz Hz) and in and Sampling is performed at intervals. On the other hand, the information symbols for each L-UE group are placed in a discrete time-frequency TF plane, which has the longest time dimension. and maximum frequency size The sampling intervals are respectively and .in and Let these represent the duration of a single symbol and the width of the subcarrier, respectively, and satisfy the following relationship: To ensure that the DD plane can fully characterize all path information of the dual-dispersion channel, and These values ​​must be greater than the channel's maximum delay spread and maximum Doppler shift, respectively. N and M represent the total number of time intervals and the number of subcarriers, respectively. To avoid fractional delay and fractional Doppler shift, N and M are set to sufficiently large values ​​in this invention. For the... Group H-UE The sequence of bits to be transmitted is constellation modulated to generate a complex data symbol sequence. .

[0063] To maximize the transmission rate of H-UE while ensuring the QoS of L-UE, the system first dynamically calculates the number of grid points that H-UE is allowed to activate in the DD domain, i.e., the sparsity, based on the current CSI. Assuming pairing The target signal-to-interference-plus-noise ratio (SINR) threshold is The power allocated to the L-UE by the base station is The equivalent channel gain of L-UE is The maximum interference power that L-UE can tolerate. for:

[0064]

[0065] in, This represents noise power.

[0066] right The interference mainly depends on the H-UE's transmit power. sparsity Because S-OTFS diffuses energy in the time-frequency domain, the effective interference power of H-UE and its sparsity ratio... It is directly proportional. Therefore, the dynamic sparsity of H-UE is... Must meet:

[0067]

[0068] in The interference channel gain from the base station to the L-UE. Interference margin protection factor (usually set to a value) The system selects those that satisfy the above conditions and The largest integer is used as the sparsity of the current transmission. .

[0069] Dynamically determine sparsity Afterwards, it is necessary to... Select on the DD domain grid Each location contains a data symbol.

[0070] In some preferred embodiments of the present invention, such as Figure 3 As shown, a sparse pattern is generated based on the sparsity. Specifically:

[0071] Based on sparsity Define the data symbol sequence index as . No. Coordinates of each symbol in the DD domain Generated by the following modulo operation rules:

[0072]

[0073] in, This represents the data symbol sequence index, where N represents the Doppler grid size and M represents the time delay dimension grid size. and Representing the Doppler step size factor and the time delay step size factor, respectively, selected as . and Large, coprime numbers are used to ensure that the generated points are uniformly scattered across the entire grid plane. and These are user-specific Doppler random offsets and time delay random offsets, used to distinguish the patterns of different H-UEs. This indicates taking the modulus. For Doppler indexing, This is a time-delay index.

[0074] Mapping H-UE symbols to the DD domain based on sparse patterns, specifically:

[0075] This invention introduces a sparse mapping matrix into the traditional OTFS, assuming... Let H-UE be the signal matrix in the DD domain. Define a sparse mask. Its elements Indicator grid points Is it activated?

[0076]

[0077] in, This represents the discrete Doppler index, corresponding to the Doppler frequency shift. , Represents a discrete time delay index, corresponding to the time delay. Therefore, a sparse mask matrix is ​​constructed based on the generated coordinate set. Its elements Defined as:

[0078]

[0079] Furthermore, the DD domain signal of H-UE The generating formula is:

[0080]

[0081] DD domain signal It needs to be converted to a time-frequency TF domain signal. This transformation is achieved through the inverse symplectic finite Fourier transform (ISFFT), and the specific formula is as follows:

[0082]

[0083] in, Indexed by time interval, For frequency domain subcarrier index, This is a normalization factor to ensure energy conservation before and after the transformation. This is used to generate the continuous-time waveform of the actual emission. TF domain signal It requires the Heisenberg transformation, that is:

[0084]

[0085] in, This is a pulse shaping filter for transmitting signals.

[0086] This invention employs a scattering-type sparse pattern generation rule. Symbol mapping using this pattern ensures that non-zero symbols in the DD domain are as far apart as possible in terms of both time delay and Doppler dimension. This maximizes the Euclidean distance between symbols, enhancing the system's ability to resist inter-symbol interference and inter-Doppler interference caused by bidispersive channels. Compared to full-set mapping, the scattering-type pattern maximizes the distance between non-zero symbols, helping to minimize inter-symbol interference (ISI) at the receiver after passing through the channel. After ISFFT transformation, the scattering-type distribution reduces the peak-to-average power ratio (PAPR) of the signal.

[0087] For the L-UE To accommodate its low-speed, low-power characteristics, SC-FDMA modulation is adopted, specifically:

[0088] L-UE data symbol sequence First of all The point-wise DFT transform converts the time-domain symbol to the frequency domain:

[0089]

[0090] in, This refers to the length of the data symbol sequence, i.e., the number of subcarriers allocated to the current L-UE user. Frequency domain symbols. Mapped to total system bandwidth Define a subcarrier mapping matrix on a specific subset of the data. .

[0091]

[0092] On the TF domain trellis, this means that the L-UE signal occupies only specific frequency indices, while being 0 at other frequency positions. Mapped frequency domain signal go through Pointed IFFT transform generates the time-domain signal, i.e., the SC-FDMA signal, and a cyclic prefix (CP) is added to eliminate inter-symbol interference (ISI).

[0093]

[0094] S4: H-UE and L-UE users receive NOMA signals and calculate the spectral efficiency of H-UE and L-UE users.

[0095] In millimeter-wave communication systems, the downlink transmission process of OTFS-NOMA technology consists of two stages: 1) beam alignment; 2) data transmission. This embodiment assumes the beamwidth of all user equipment. A fixed beam can cover the entire communication range, while the base station needs to search for the optimal transmission beam across the entire beam space. In the first phase, this search is performed sequentially for each user equipment. Therefore, the required search time is... It is given by the following formula:

[0096]

[0097] in, This indicates the time required for the pilot signal to transmit. It is a fixed value representing the coverage area of ​​the base station sector. and These represent the base station sending to the first Group 1 The H-UE user and the first Group 1 The beamwidth transmitted by each L-UE user.

[0098] Based on the characteristics of millimeter-wave communication, the effective antenna gain expressions for the main lobe of the receiver and transmitter of H-UE are as follows:

[0099]

[0100]

[0101] in, This represents the sidelobe gain, assuming the same sidelobe gain at the base station transmitter and the user receiver (H-UEs / L-UEs). Additionally, it represents the effective antenna gain of the main lobe at the L-UE transmitter. Similar to the above formula.

[0102] In millimeter-wave NOMA systems, the first The first in the group After an H-UE completes OTFS demodulation in the DD domain, its received signal can be represented as:

[0103]

[0104] in, and The initial signal is generated by S-OTFS modulation for H-UE users and SC-FDMA modulation for L-UE users. This represents the total gain of the H-UE. This represents background white noise, which follows a Gaussian distribution. And all H-UE / L-UE have the same variance , The channel matrix is ​​obtained by cyclically shifting each row of the identity matrix through the channel response derived in S1.

[0105]

[0106] in, Indicates the first Group H-UE The first H-UE Sub-bands are derived from the channel response. and They represent the first Path The discrete Doppler tap index and the time delay tap index are specifically derived from the following formula:

[0107]

[0108] Assuming that all DD domain resource blocks / TF domain resource blocks carrying information symbols in a single H-UE / L-UE have the same power, that is, the transmit power of H-UE and L-UE can be defined as follows: and Therefore, the second and third terms on the right-hand side of the formula for the received signal of the H-UE user group are the interference noise terms, and their covariance matrix is ​​shown below:

[0109]

[0110] Therefore, the spectral efficiency is given by the following formula:

[0111]

[0112] in, This represents the average power of L-UE users. Indicates the first Group H-UE Spectral efficiency for individual users. This indicates the number of users in each user group. Indicates the total gain of H-UE. This indicates the effective antenna gain of the main lobe at the receiving end. This indicates the effective antenna gain of the main lobe at the transmitting end; Indicates that the base station sends to the first Group H-UE The power transmitted by each user, Indicates the number of L-UE user groups. Indicates the first Group H-UE and the first Pairing variables for L-UE pairing.

[0113] When using Non-Orthogonal Multiple Access (NOMA) technology, the base station... Each SC-FDMA subcarrier and Several L-UE users communicate, and these carriers are simultaneously shared with H-UEs. Assuming that all H-UE signals can be eliminated by serial interference cancellation (SIC) technology, and therefore the signal is only affected by time-invariant channel gain and only by frequency-selective fading, then the... In group L-UE, the first L-UE users The specific expression for the sub-band is:

[0114]

[0115] in, Represents the noise term, and To simplify the analysis, it is assumed that all users have the same noise variance. This indicates that the signal is in Given the time-frequency plane channel gain in the carrier band, the spectral efficiency of the signal can be expressed as:

[0116]

[0117] in, and .in It is an L-UE index within the group. It is occupied by The index in the resource block, where W represents the number of TF resource blocks occupied by each L-UE. Indicates the first Group 1 A signal in Time-frequency plane channel gain on the carrier band, Indicates the first Group 1 L-UE transmit power. This represents the total gain of L-UE.

[0118] S5: Construct an optimization problem to maximize the average spectral efficiency of H-UE based on the spectral efficiency of H-UE users and L-UE users.

[0119] For the constructed S-OTFS-NOMA millimeter-wave downlink communication system, this embodiment considers the joint optimization problem of transmit power, beamwidth, and user equipment scheduling. Under the conditions of satisfying the minimum throughput constraint for L-UE, total time constraint, transmit power constraint, beamwidth constraint, and user equipment scheduling constraint, this embodiment aims to maximize the average spectral efficiency of H-UE. The original optimization problem can be expressed as follows:

[0120]

[0121] in, , and Representing the sets of power variables respectively Beamwidth set and user equipment scheduling variable set . To achieve the minimum achievable beamwidth, This represents the maximum achievable beamwidth. It is the total duration of beam alignment and communication. This represents the minimum throughput required for each L-UE. Indicates that the base station sends to the first Group H-UE The power of the signal transmitted by each user. Indicates that the base station sends to the first The first L-UE The power of the signal transmitted by each user. Indicates the first The average power of the L-UE group This indicates the maximum transmit power of the base station. Indicates beam alignment time; Indicates the first Group H-UE Spectral efficiency for individual users. This indicates the number of TF resource blocks occupied by each L-UE, i.e. , Indicates the first The NOMA group used the first Scheduling variables for each frequency band Indicates the first Group H-UE and the first Pairing variables for L-UE pairings, Indicates the first In group L-UE, the first The user in the first Spectral efficiency across frequency bands Indicates the subcarrier spacing. Indicates the first The average transmit power of group L-UE, Indicates the first Group 1 The transmit power of each H-UE Indicates the number of H-UE user groups. Indicates the number of frequency band indices. Indicates the number of L-UE user groups. This indicates the number of users in each user group. Represents the H-UE user group index set. Represents the L-UE user group index set. Represents a set of frequency band indices. This represents the set of user indexes within each user group.

[0122] In high-mobility scenarios, H-UEs are susceptible to severe Doppler shift, preventing them from prioritizing the decoding of L-UE signals. To address cross-domain interference, H-UEs employ higher transmit power and reduce interference to L-UEs through sparse mapping to achieve priority decoding. L-UEs, in turn, eliminate H-UE interference using SIC technology. Therefore, constraint C3 ensures that H-UE signals can be decoded preferentially; C4 specifies that beam alignment time must be less than the total duration; C5 defines the available beamwidth range; C6 indicates a one-to-one mapping between each NOMA group and each available frequency band; C7 explains a one-to-one mapping between each H-UE group and each L-UE group; and C8 indicates that the scheduling variable is a Boolean variable.

[0123] S6: Solve the optimization problem of maximizing the average spectral efficiency of H-UE to obtain the resource allocation scheme; the system transmits signals according to the resource allocation scheme.

[0124] As can be seen from the H-UE and L-UE spectral efficiency formulas, there is a coupling relationship between transmit power, beamwidth, and scheduling variables. Based on these existing nonlinear expressions, neither the objective function nor the constraint C1 possesses convexity. Furthermore, the scheduling variables in the three-way matching problem, as discontinuous binary values, are also coupled with each other. The constructed optimization problem is a non-convex and difficult-to-handle combinatorial problem, and since the scheduling variables are not directly related to beamwidth and transmit power, this embodiment proposes an alternating optimization-multi-round relaxation AO-MRA scheme to decouple the original problem, which is a mixed integer non-convex optimization problem, into two tractable subproblems. That is, the original joint scheduling, beamwidth, and transmit power allocation problem is decoupled into a scheduling subproblem P1 and a beam-power joint allocation subproblem P2. The method of solving P1 and P2 is adopted by combining alternating optimization and multi-round relaxation. Specifically, under the condition of fixing some variables, the other part of the variables is optimized, and through variable relaxation, introduction of relaxed variables and continuous convex approximation method, the non-convex constraints and nonlinear objective function are transformed into a solvable convex approximation problem, thereby obtaining the stage optimal solution.

[0125] The process involves alternating iterative updates between the two sub-problems. In each iteration, a convex approximation model is reconstructed and solved based on the current solution until the objective function value or variable change satisfies the preset convergence condition. Finally, the obtained continuous solutions undergo necessary binarization or quantization to determine the frequency band scheduling results, beamwidth, and transmit power allocation scheme for each NOMA group, and based on this, the downlink signal resource allocation and transmission implementation are completed. Specifically:

[0126] S61: With fixed power and beamwidth, the optimization problem is modeled as a non-convex nonlinear programming (INLP) problem P1. This non-convex optimization problem P1 is effectively solved by the continuous convex approximation (SCA) method.

[0127] Given the transmit power and beamwidth satisfying the constraints C2 and C5 in the initial problem, and after relaxing the scheduling variables in C8, a scheduling subproblem P1 of the following form is constructed:

[0128]

[0129] in, and Clearly, C6, C7, and C8 in the original objective function are linear constraints. However, the fractional form of P1 and the coupling relationship between the scheduling variables in C1 of the original objective function make subproblem P1 a non-convex optimization problem. To handle the nonlinear expression in the objective function P1, slack variables are introduced. and To address the issues in C1 of the original objective, slack variables were also introduced. ,in , , Therefore, P1 can be equivalently represented as P1.1:

[0130]

[0131] in, , and Representing all slack variables , and In the set, C3 in P1.1 is a constraint condition. norm form, This represents the communication time. Through this transformation, the objective function in P1 becomes a concave function in P1.1, with C3 and C4 being convex constraints. However, due to the presence of fractional variables... In P1.1, C1 is a non-convex constraint. Furthermore, due to the presence of second-order variables, C2 is also a non-convex constraint. Therefore, P1.1 remains a non-convex optimization problem.

[0132] For fractional variables Its lower bound can be obtained through the corresponding first-order Taylor expansion, that is:

[0133]

[0134] in, This indicates the current iteration point.

[0135] Non-convex constraints are replaced with convex constraints:

[0136]

[0137] For C2 in P1.1, it needs to be processed. Similarly at the iteration point For convex functions Approximate representation using a first-order Taylor expansion:

[0138]

[0139] In P1.1, C2 can be replaced with a linear constraint:

[0140]

[0141] Thus, P1.1 can be approximated as a convex problem, and can then be solved iteratively using the following scheduling algorithm.

[0142] The principle of the proposed scheduling algorithm is summarized in Algorithm 1. First, the parameters are determined based on the relevant constraints of problem P1.1. , and The initial value is then determined. Subsequently, the optimal solution to problem P1.1 is obtained using the CVX toolbox in each iteration. , , and The initial parameters are updated using the optimal solution from the current iteration to prepare for the next iteration. The optimal solution is obtained iteratively. , And update the initial value , and This continues until the objective function converges, ultimately yielding the solution to problem P1. and .

[0143] S62: With the scheduling result fixed, the optimization problem is transformed into a non-convex continuous variable optimization problem P2 with respect to beamwidth and transmit power.

[0144] Based on the scheduling variables obtained in S61, the optimization subproblem P2 is constructed as follows:

[0145]

[0146] Clearly, constraints C2 to C5 in P2 are all convex constraints. However, since the subproblem of P2 and C1 both exist in fractional form... and This problem belongs to the category of nonconvex optimization problems. Furthermore, beam-related... Also related to spectral efficiency The existence of coupling relationships gives P2 a highly nonlinear characteristic. To make P2 solvable, the non-convex objective function and constraints need to be transformed into a convex form.

[0147] To address the complex optimization objective in P2, its positive and negative parts are treated separately. To remove fractional coupling and facilitate convex approximation of SCA, an inverse power variable is introduced. and Furthermore, slack variables are introduced to simplify the objective function expression. and Therefore, the positive part of the objective function can be rewritten as:

[0148]

[0149] in, , and They represent the sets of slack variables respectively. and , , , This refers to other users within the same group. Set of equivalent variables Note the optimization objective in the above formula. This results in the objective being a non-convex problem. , , and This also makes the problem non-convex. Similarly, a first-order Taylor approximation can be used to obtain the upper or lower bound of the above function. For a given set of feasible points... The following inequality must be satisfied:

[0150]

[0151] in, express The upper realm.

[0152] For a given set of feasible points and The lower realm and It can be represented as:

[0153]

[0154]

[0155] For a given set of feasible points The lower realm and It can be represented as:

[0156]

[0157]

[0158] For the negative part of the objective function, the same convex approximation SCA method is used, thus transforming the objective function P2 into a convex optimization problem.

[0159] To resolve the non-convex constraint C1 in P2, slack variables are introduced. Rewrite it as:

[0160]

[0161]

[0162] in, However, the above equation is still a non-convex constraint, therefore for a given set of points... and Its lower boundary It can be represented as:

[0163]

[0164] Based on the above steps, P2 can be transformed into a convex problem. The parameters are initialized according to the relevant constraints. The optimal solution of the slack variables is obtained by solving P2 using the CVX toolbox. The slack variables are then used to solve P2 suboptimally using Algorithm 2.

[0165]

[0166] In summary, this invention utilizes the discretized sparse pattern of sparse OTFS to perform sparse mapping of high-speed user signals in the DD domain, and extends it to the entire TF domain through inverse symplectic fast Fourier transform, thereby significantly reducing the power spectral density per unit frequency band and the interference level to low-speed users in the same frequency band. This invention introduces sparse mapping OTFS to construct a resource allocation optimization model. Under the premise of ensuring QoS for low-speed users, it solves the problem through an improved AO-MRA strategy, decoupling the non-convex mixed integer programming problem into a user scheduling subproblem and a joint beam and power allocation subproblem, and iteratively solving it using SCA technology. This invention reduces interference noise floor at the physical layer through sparse waveform design, thereby effectively releasing the power margin of high-speed users while keeping the total system power constraint unchanged, significantly improving the communication system, rate, and spectral efficiency in heterogeneous mobility scenarios.

[0167] The above-described embodiments further illustrate the purpose, technical solution, and advantages of the present invention. It should be understood that the above-described embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made to the present invention within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, characterized in that, include: S1: Construct a NOMA system model based on S-OTFS; S2: Cluster and pair high-speed mobile users (H-UEs) with low-speed mobile users (L-UEs) to obtain multiple NOMA groups; allocate subbands for each NOMA group; S3: Generate a sparse pattern and map the H-UE symbols to the DD domain according to the sparse pattern, and generate a sparse OTFS signal after ISFFT processing; map the L-UE symbols to the TF domain and perform SC-FDMA modulation to generate an SC-FDMA signal; The base station transmits a NOMA signal composed of OTFS and SC-FDMA signals; S4: H-UE users and L-UE users receive NOMA signals and calculate the spectral efficiency of H-UE users and L-UE users; S5: Construct an optimization problem to maximize the average spectral efficiency of H-UE based on the spectral efficiency of H-UE users and L-UE users; S6: Solve the optimization problem of maximizing the average spectral efficiency of H-UE to obtain the resource allocation scheme; the system transmits signals according to the resource allocation scheme.

2. The resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The NOMA system model based on S-OTFS includes a base station (BS) and... Users are categorized into high-speed mobile users (H-UE) and low-speed mobile users (L-UE) based on their mobile speed. H-UE uses sparse OTFS modulation for its data symbols, while L-UE uses SC-FDMA modulation for its data symbols.

3. The resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The process of clustering and pairing high-speed mobile user (H-UE) and low-speed mobile user (L-UE) includes: clustering an H-UE group and an L-UE group into a NOMA group, and each NOMA group contains H-UE users and L-UE users.

4. The resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The process of generating a sparse pattern and mapping H-UE symbols to the DD domain based on the sparse pattern includes: Determine the sparsity of the current transmission; Define a data symbol sequence index based on sparsity; generate a sparse pattern based on the data symbol sequence index; Construct a sparse mask matrix based on the sparse pattern; The H-UE symbols are mapped to the DD field based on the sparse mask matrix.

5. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 4, is characterized in that... The sparsity is the largest integer that satisfies the condition, which is expressed as: ； ； Where K represents sparsity. This indicates the interference margin protection factor. This represents the interference channel gain from the base station to the L-UE. This indicates the maximum interference power that the L-UE can tolerate. The value represents the transmit power of the H-UE, N represents the Doppler grid size, and M represents the time delay dimension grid size.

6. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 4, is characterized in that... The sparse pattern representation generated based on the data symbol sequence index is as follows: ； in, This represents the data symbol sequence index, N represents the Doppler grid size, and M represents the time delay dimension grid size. Indicates Doppler index, Indicates a delay index. and These represent the Doppler step size factor and the time delay step size factor, respectively. and These are represented as Doppler random offset and time delay random offset, respectively. This indicates taking the modulus.

7. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The formula for calculating the spectral efficiency of H-UE users and L-UE users is as follows: ； ； in, Indicates the first Group H-UE Spectral efficiency for individual users. Indicates the first In group L-UE, the first The user in the first Spectral efficiency across frequency bands W represents the number of users in each user group, and W represents the number of TF resource blocks occupied by each L-UE. Indicates the total gain of H-UE. This represents the total gain of L-UE. Indicates that the base station sends to the first Group H-UE The power of the signal transmitted by each user. Indicates the number of L-UE user groups. Indicates the first Group H-UE and the first Pairing variables for L-UE pairings, Indicates the first In group L-UE, the first Individual users Time-frequency plane channel gain on the carrier band, Indicates that the base station sends to the first Group 1 The power of the transmitted signal of each L-UE user, Indicates the first The average transmit power of group L-UE, This represents the noise variance.

8. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The optimization problem of maximizing the average spectral efficiency of H-UE is expressed as: ； in, , and These represent the power variable set, the beamwidth set, and the user equipment scheduling variable set, respectively. Indicates the total duration of beam alignment and communication. This indicates the beam alignment time, and W represents the number of TF resource blocks occupied by each L-UE. Indicates the first Group H-UE Spectral efficiency for individual users. Indicates the subcarrier spacing. This indicates the maximum transmit power of the base station. Indicates that the base station sends to the first The first L-UE The power of the signal transmitted by each user. Indicates the first The average transmit power of group L-UE, Indicates that the base station sends to the first Group H-UE The power of the signal transmitted by each user. This represents the minimum achievable beamwidth. This indicates the maximum achievable beamwidth. Indicates the first The NOMA group used the first Scheduling variables for each frequency band Indicates the first Group H-UE and the first Pairing variables for L-UE pairings, Indicates the first In group L-UE, the first The user in the first Spectral efficiency across frequency bands This represents the minimum throughput required for each L-UE. Indicates the number of H-UE user groups. Indicates the number of frequency band indices. Indicates the number of L-UE user groups. This indicates the number of users in each user group. Represents the H-UE user group index set. Represents the L-UE user group index set. Represents a set of frequency band indices. This represents the set of user indexes within each user group.

9. A resource allocation method for a network where high-speed mobile terminals and low-speed users coexist, as described in claim 1, is characterized in that... The process of solving the optimization problem of maximizing the average spectral efficiency of H-UE includes: The optimization problem of maximizing the average spectral efficiency of H-UE is decoupled into a scheduling subproblem and a beam-power joint allocation subproblem. Under the condition of fixing some variables, the other part of the variables is optimized. By using variable relaxation, relaxation variable introduction and continuous convex approximation methods, the non-convex constraints and nonlinear objective functions of the two subproblems are transformed into convex approximation problems. The convex approximation problems are solved to obtain the stage-optimal solution. Alternating iterative updates are performed between the two sub-problems. In each iteration, a convex approximation model is reconstructed and solved based on the current solution until the objective function value or variable change satisfies the preset convergence condition. Finally, the obtained continuous solution is binarized or quantized to determine the frequency band scheduling results, beamwidth, and transmit power allocation scheme of each NOMA group, i.e., the resource allocation scheme.

Images