A semi-positive definite resource allocation method for Internet of Vehicles based on network slicing and NOMA
By adopting the static and dynamic spectrum resource pool allocation method of network slices and NOMA in the Internet of Vehicles, and combining iterative algorithms to optimize the transmission power of V2V users, the problems of low resource allocation efficiency and poor spectrum benefits in the Internet of Vehicles are solved, and high efficiency is achieved resource utilization and fair distribution.
Patent Information
- Application Number
- CN202210729292.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-24
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2042-06-24
AI Technical Summary
The prior art has insufficient research on resource allocation based on network slices and NOMA in the Internet of Vehicles, which is difficult to meet the diversified needs of vehicle users for data transmission rates and delays, and the resource allocation efficiency is low and the spectrum benefits are poor.
Using the static and dynamic spectrum resource pool allocation method based on network slices, V2I users allocate time-frequency resource blocks, and optimize the transmission power of V2V users through iterative algorithms to meet the QoS indicators and delay requirements of each user group, and improve the Resource utilization rate.
The transmission energy efficiency and spectrum utilization of V2I users in the Internet of Vehicles have been improved, the computational complexity and time of resource allocation are reduced, and the fairness and efficiency of resource allocation are ensured.
Smart Images
Figure CN115209554B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of wireless communication technology, and in particular to a semi-positive definite resource allocation method for an Internet of Vehicles based on network slicing and NOMA (non-orthogonal multiple access). Background Art
[0002] The different requirements of V2X (vehicle-to-everything) communication in the Internet of Vehicles for the data transmission rate, transmission delay and reliability of the network will cause suboptimal utilization of 5G networks. The differences in the types of vehicle user services in the Internet of Vehicles put forward different requirements for network resources. For example, services such as media streaming and information exchange require a large amount of data exchange, while navigation services are delay-sensitive. At the same time, with the rapid development of the Internet of Vehicles, more and more vehicle users have the need to access the network, and the scale of network space interfaces is challenged. It is necessary to study the application of network slicing based on SDN (software defined network) and NFV (Network Functions Virtualization) and NOMA resource allocation in the Internet of Vehicles.
[0003] The high-speed mobility of vehicle users makes centralized resource allocation that relies on obtaining user CSI (channel state information) no longer applicable, and various service types have differentiated requirements for network resources. In addition, there are still problems in the Internet of Vehicles, such as tight air interface resources and low spectrum efficiency. Therefore, it is necessary to develop a semi-positive definite resource allocation method for the Internet of Vehicles based on network slicing and NOMA. Most of the existing literature studies on network slicing-based resource allocation estimate the channel of vehicle users on the premise that the base station knows the user's CSI, and the resource allocation optimization methods are mainly exhaustive search, greedy algorithm, etc. NOMA has potential in improving spectrum efficiency and energy efficiency. The existing NOMA-based Internet of Vehicles resource allocation combined with network slicing is still understudied, and most of them use machine learning and neural networks as optimization methods. Finding the optimal solution requires a large number of trainings, which is not efficient. Therefore, the application of network slicing and NOMA in Internet of Vehicles resource allocation still needs to be fully studied. Summary of the invention
[0004] Technical problem: The purpose of the present invention is to provide a semi-positive definite resource allocation method for Internet of Vehicles based on network slicing and NOMA to solve the above-mentioned problem.
[0005] Technical solution: To solve the above technical problems, a semi-positive definite resource allocation method for Internet of Vehicles based on network slicing and NOMA is specifically as follows:
[0006] Step 1: Vehicle users communicating with infrastructure are referred to as V2I users. It is assumed that the V2I user group in the same network slice is referred to as a V2I slice group. When the number of V2I slice groups and the transmission rate of the user communication quality QoS indicator in each V2I slice group are known, the static spectrum resource pool is used to initialize the time-frequency resource block RB allocation for each V2I slice group. Furthermore, a semi-positive definite resource allocation time period is referred to as an SPS (semi-persistent scheduling) period, and an SPS period includes a resource reservation management phase and a data transmission phase. A minimum granularity of allocable time-frequency resources is referred to as a time-frequency resource block RB. The base station uses a dynamic spectrum resource pool to fine-tune the time-frequency resource block RB allocation of each V2I slice group according to the downlink data transmission rate fed back by each V2I slice group in the data transmission phase of the previous SPS period, until the QoS indicator transmission rate of each V2I slice group is met. If there are still excess time-frequency resource blocks RB that have not been allocated, the remaining sub-channels are allocated one by one with different probabilities according to the data transmission rate of each V2I slice group until they are fully allocated.
[0007] Step 2: The group of transmitting users of vehicle-to-vehicle communication that are divided into the same non-orthogonal multiple access cluster is recorded as a V2V NOMA cluster, and the vehicle users of vehicle-to-vehicle communication are recorded as V2V users; the transmitting users of vehicle-to-vehicle communication are recorded as V2V Tx users; the receiving users of vehicle-to-vehicle communication are recorded as V2V Rx (receiver) users; in the data transmission phase of the current SPS cycle, assuming that the spectrum sharing mode of the V2V NOMA cluster and each V2I slice group and the V2V Tx users in each V2V NOMA cluster are known, it is stipulated that V2V Rx receives the broadcast signal of V2V Tx within a specific receiving range; V2V Tx users and V2V Rx users within the receiving range perform distributed transmission power control updates on V2V Tx users through an iterative algorithm, so as to improve the transmission energy efficiency of V2V users while meeting the delay requirements.
[0008] in,
[0009] The step 1 is specifically as follows:
[0010] Step 1.1: Assume that the transmission power of the base station on each time-frequency resource block RB is fixed, and assume that there are M V2I users within the coverage area of the base station. The mth V2I user is denoted as V2I m. The QoS indicator transmission rate of each V2I user is denoted as The number of V2I slice groups is recorded as S, and the V2I user and V2I slice group belonging discrimination matrix Affi_NS of M×S dimension is defined, Affi_NS(m,s)=1 indicates that V2I m belongs to V2I slice group s, otherwise Affi_NS(m,s)=0;
[0011] Define an S-dimensional vector aveQoSrate, where the sth element represents the average QoS indicator transmission rate of V2I slice group s, that is:
[0012]
[0013] Step 1.2: The allocatable time-frequency resources in the base station are divided into a static time-frequency resource pool and a dynamic time-frequency resource pool. According to the vector aveQoSrate in step 1.1, the time-frequency resource block RB ratio of the static spectrum resource pool of each V2I slice group is determined. The static time-frequency resource block ratio of V2I slice group s is recorded as
[0014]
[0015] The total bandwidth of the time-frequency resource blocks in the static spectrum resource pool is Then the bandwidth allocated by V2I slice group s from the static spectrum resource pool in the initialization phase is
[0016] Step 1.3: The actual transmission power of V2Im is Each V2I user uploads the downlink transmission rate of the previous SPS cycle through the SPS cycle uplink The base station collects statistics on the QoS indicator transmission rate satisfaction of each V2I slice group, defines an S-dimensional QoS indicator transmission rate satisfaction discriminant vector meetQoS, initializes it to all zeros, and the sth element can represent the satisfaction of the QoS indicator transmission rate of the V2I slice group s, that is:
[0017]
[0018]
[0019] Where sgn(x) is a sign function, when x>0, sgn(x)=1; when x=0, sgn(x)=0; when x<0, sgn(x=-1; if meetQoS(s)=1, the QoS indicator transmission rate of all V2I users in V2I slice group s is met, otherwise the QoS indicator transmission rate of some V2I users in V2I slice group s is not met;
[0020] Step 1.4: If for any V2I slice group s, meetQoS(s) in the previous SPS cycle is less than 1, allocate RBs from the dynamic spectrum resource pool to the V2I slice group s at the minimum granularity, i.e., a single time-frequency resource block RB, and repeat step 1.4 until meetQoS is a vector of all 1s in the current SPS cycle;
[0021] Step 1.5, set the S-dimensional dynamic allocation variable μ. If there are still idle time-frequency resource blocks RB in the dynamic spectrum resource pool, then for each idle time-frequency resource block RB, in order to properly consider fairness, it tends to be allocated to the V2I slice group that can maximize the μ value in the current SPS period, with probability The time-frequency resource blocks RB are allocated to the V2I slice group s, where:
[0022]
[0023] Where |RB s | represents the number of time-frequency resource blocks (RBs) of V2I slice group s, RB s Represents the time-frequency resource block RB set of the V2I slice group s; repeat step 1.5 until there are no idle time-frequency resource blocks RB in the dynamic spectrum resource pool.
[0024] The step 2 is specifically as follows:
[0025] Step 2.1: Assume that there are J V2V NOMA clusters, P V2V Tx users, and Q V2V Rx users within the coverage area of the base station. Define a P×J-dimensional matrix Affi_NOMA, where Affi_NOMA(p,j)=1 indicates that V2V Txp belongs to V2V NOMA cluster j; otherwise, Affi_NOMA(p,j)=0; the set of time-frequency resource blocks in V2V NOMA cluster j is RB j ,The power control of V2V Tx in a single V2V NOMA cluster is constructed as follows.,Taking V2V NOMA cluster j as an example, the optimization problem P * The build is as follows:
[0026]
[0027]
[0028]
[0029]
[0030]
[0031]
[0032]
[0033]
[0034] Optimization Problem P * It contains five constraints C1 to C5. The pth V2V Tx user is V2V Tx p; the qth V2V Rx user is V2V Rx q; the nth time-frequency resource block RB is RB n. represents the transmit power of V2V Tx p on RBn, is the channel gain of V2V Tx p and V2V Rx q on RB n, is the channel gain from the base station to V2V Rx q on RB n, The channel gain on RB n is weaker than The channel gain between V2V Tx p′ and V2V Rx q. is the downlink power of the V2I user that is the nearest neighbor to V2V Rx q and shares spectrum resources with V2V NOMA cluster j; is the maximum transmission power of each V2V Tx user, is the transmission rate of V2V Tx p and V2V Rx q on RB n, R pq is the transmission rate between V2V Tx p and V2V Rx q, L is the broadcast packet length, D is the maximum tolerable delay, B RB is the bandwidth of a single RB. 2 is the noise power;
[0035] Step 2.2: The communication distance between V2V users is recorded as D Tx , then in the V2V Tx p radius D Tx All V2V Rx users within the range can receive the broadcast data packets of V2V Tx p. Therefore, only the V2V Rx users and V2V Tx p that can receive the V2V Tx p signal are considered to jointly optimize the power control of V2V Tx p. If the V2V Tx p radius D Tx There is only one V2V Rx q within the radius D of the V2V Rx q. Tx If there is no other V2V Tx user in the same V2V NOMA cluster as V2V Tx p within the range, set the temporary power variable Temp1_P p ,Temp2_P p ;
[0036] Step 2.3: If the radius D of V2V Tx p TxThere is only one V2V Rx q within the radius D of the V2V Rx q. Tx If there are other V2V Tx users in the same V2I slice group as V2V Tx p within the range, the rotation strategy is adopted to optimize the radius D of the V2V Rxq Tx Power control of each V2V Tx p within;
[0037] Step 2.4: If the radius D of V2V Tx p Tx Inside, there is L p V2V Rx, define the counting variable l, then this L p V2VRx is denoted as q l Through this L p Each V2V Rx jointly optimizes the transmission power of the V2V Tx p.
[0038] The step 2.2 is specifically as follows:
[0039] Step 2.2.1, Initialization state, Temp1_P p =0, Note | RB j | is the number of time-frequency resource blocks (RBs) of V2V NOMA cluster j, and is taken as:
[0040]
[0041] Step 2.2.2: According to the optimization problem P * Based on the constraints C2 and C3, calculate the transmission rate R of V2V Rx q and V2V Tx p pq ;
[0042] Step 2.2.3: If Update like Update
[0043] Step 2.2.4, set the error precision ∈, if |Temp1_P p -Temp2_P p |>∈, repeat steps 2.2.1 to 2.2.3 until |Temp1_P p -Temp2_P p |<∈, the total transmission power of V2V Tx p is finally updated as And take As the transmission power of V2V Tx p on RB n.
[0044] The step 2.3 is specifically as follows:
[0045] Step 2.3.1. Assume that the radius of V2V Rx q is D Tx There is K j The V2V Tx users of the V2V NOMAj cluster have j The channel gains between V2V Tx users and V2V Rx q are arranged in descending order, that is, For this K j V2V Tx user settings 2K j A temporary power variable, denoted as Where k is a counting variable,
[0046] Step 2.3.2: initialization Pick:
[0047]
[0048] Step 2.3.3, k = 1, according to the constraints C2 and C3, without changing other K j -1 V2V Tx user transmission power, calculate the transmission rate R of V2V Rx q and V2V Tx p pq ;
[0049] Step 2.3.4: If Update like Update renew As the transmission power of V2V Tx p on RB n, the total transmission power of V2V Tx p is
[0050] Step 2.3.5, k = k + 1, repeat steps 2.3.3 to 2.3.4, and repeat this step until K j V2V Tx users are traversed;
[0051] Step 2.3.6: Take steps 4.4.2 to 4.4.5 as one cycle, and repeat multiple cycles until End the loop.
[0052] The step 2.4 is specifically as follows:
[0053] Step 2.4.1. Define L for V2V Tx p p A temporary power optimization variable, denoted as P pl , initialization Set the iteration round count variable t and initialize t=0;
[0054] Furthermore, for the optimization problem P * The objective function is modified to:
[0055]
[0056] And add new constraints:
[0057]
[0058] Construct the augmented Lagrangian function:
[0059]
[0060] where v l is the Lagrange multiplier, initialized to v l (t), is the penalty term; the above formula is equivalent to:
[0061]
[0062] Step 2.4.2: When l = 1, through P pl With V2V Rx q l The joint power optimization of V2V Rx q l The radius D Tx If there are no other V2V Tx users in the same V2V NOMA cluster as V2V Tx p within the range, then implement steps 2.2.1 to 2.2.4. After If the update result is reduced, keep it, otherwise repeat steps 2.2.1 to 2.2.4 to find the one that makes Reduced and update
[0063] Step 2.4.3, if in the V2V Rx q l The radius D Tx If there are other V2V Tx users in the same V2I slice group as V2V Tx p within the range, then implement steps 2.3.1 to 2.3.5. After If the update result is reduced, keep it, otherwise repeat steps 2.3.1 to 2.3.5 to find the one that makes Reduced and update
[0064] Step 2.4.4, Update Make According to the principle of least squares, we know
[0065] Step 2.4.5, update the Lagrange multiplier:
[0066]
[0067] Step 2.4.6, l=l+1, repeat steps 2.4.1 to 2.4.5 until L temporary power optimization variables are traversed;
[0068] Step 2.4.7: Take steps 2.4.1 to 2.4.6 as one round, repeat multiple rounds, define intensive reading ∈ 2, until L equal If the decrease is less than ∈2, stop updating and take The power allocation result.
[0069] Table 1 Index of main symbols in this paper
[0070]
[0071]
[0072]
[0073] Beneficial effects: Compared with the prior art, the present invention is improved in the following aspects:
[0074] (1) In view of the fact that it is difficult to rely on obtaining the CSI of vehicle users to implement resource allocation based on network slicing in the Internet of Vehicles, among V2I users, considering the QoS requirements of V2I users for data transmission rate, the allocation of V2I slice group time-frequency resource blocks (RBs) based on the QoS satisfaction in continuous SPS cycles is implemented for each V2I slice group within the coverage of the base station.
[0075] (2) In order to meet the QoS indicator transmission rate requirements of V2I users in each V2I slice group in the shortest possible time, the spectrum resource pool of the base station is divided into two parts: a static spectrum resource pool and a dynamic spectrum resource pool. The static spectrum resource pool can initialize the allocation of time-frequency resource blocks (RBs) for each V2I slice group. After initialization, the base station uses the dynamic spectrum resource pool to fine-tune the allocation of time-frequency resource blocks (RBs) for each V2I slice group based on the satisfaction of the QoS indicator transmission rate of each V2I slice group. After ensuring the QoS of each V2I slice group, the remaining resources in the static spectrum resource pool are allocated fairly to avoid the situation where individual V2I slice groups occupy the vast majority of time-frequency resource blocks (RBs). The allocation of time-frequency resource blocks (RBs) for the entire V2I slice group not only ensures the transmission energy efficiency of V2I users, but also takes into account the fairness of spectrum resource allocation for the V2I slice group.
[0076] (3) In the transmission power control optimization process of V2V Tx based on NOMA, considering the fast time-varying characteristics of vehicle user channels in the IoV scenario, it is difficult for the base station to obtain the CSI of V2V users to implement centralized resource allocation management. At the same time, considering the actual transmission of broadcast signals in the IoV, a reasonable V2V user communication distance D is set. Tx , the power control of V2VTx is effectively implemented through a distributed iterative algorithm. In addition, based on the V2V user communication distance D Tx , the number of V2V Rx users within the transmission range of each V2V Tx user is discussed. There is only one V2V Rx user within the transmission range of the V2V Tx user and the radius D of the V2V Rx user Tx If there is no other V2V Tx in the same cluster as the V2V Tx, the binary search method is used to search for the optimal solution that meets the constraints; if there is only one V2V Rx user within the transmission range of the V2V Tx user and the radius D of the V2V Rx user is Tx If there is a V2V Tx in the same cluster as the V2V Tx, a rotation strategy is used to optimize the power control of the V2V Tx; if there are multiple V2V Rx in the transmission range of the V2V Tx user, multiple auxiliary variables are used and a penalty term is introduced in the Lagrangian function to finally achieve the power control of the V2V Tx. This method has a small number of iterations and high computational efficiency.
[0077] Based on the above improvements, the algorithm proposed in the present invention can effectively meet the QoS indicator transmission rate of V2I while using NOMA among V2V users to obtain the following greater benefits: first, the computing speed can be greatly improved; second, the spectrum utilization rate can be effectively improved; third, the system transmission energy efficiency can be greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0078] Figure 1 This is a schematic diagram of the optimization scenario of resource allocation on straight roads in the Internet of Vehicles based on network slicing and NOMA;
[0079] Figure 2 This is a schematic diagram of the Internet of Vehicles resource allocation process based on network slicing and NOMA. DETAILED DESCRIPTION
[0080] The present invention provides a semi-positive definite resource allocation method for the Internet of Vehicles based on network slicing and NOMA. First, the base station initializes the RB time-frequency resource block allocation of each V2I slice group according to the QoS indicator transmission rate of each V2I slice group using the positive definite spectrum resource pool. At the same time, the base station adjusts the frequency resource block RB allocation of each V2I slice group in the current SPS cycle according to the data transmission rate of each V2I slice group in the previous SPS cycle to maximize the transmission energy efficiency of V2I users.
[0081] For example, there are 3 V2I slice groups within the coverage of the base station, that is, S=3, and the V2I users in each V2I slice group have their own QoS indicator transmission rate. There are 9 V2I users within the coverage of the base station. For the convenience of annotation, it is assumed that:
[0082] Affi_NS(1,1)=1; Affi_NS(2,1)=1; Affi_NS(3,1)=1;
[0083] That is, when m=1, 2, 3, the corresponding V2I 1, V2I 2, V2I 3 belong to V2I slice group 1 (s=1);
[0084] Affi_NS(4,2)=1; Affi_NS(5,2)=1; Affi_NS(6,2)=1;
[0085] That is, when m=4, 5, 6, the corresponding V2I 4, V2I 5, V2I 6 belong to V2I slice group 2 (s=2);
[0086] Affi_NS(7,3)=1; Affi_NS(8,3)=1; Affi_NS(9,3)=1;
[0087] When m=7, 8, 9, the corresponding V2I 7, V2I 8, V2I 9 belong to V2I slice group 3 (s=3).
[0088] Assume that the QoS indicator transmission rate of the V2I user in V2I slice group 1 is Assume that the QoS indicator transmission rate of the V2I user in V2I slice group 2 is Assume that the QoS indicator transmission rate of the V2I user in V2I slice group 3 is According to step 1.1, the average QoS indicator transmission rate of the three V2I slice groups is:
[0089]
[0090]
[0091]
[0092] According to step 1.2, the bandwidth allocated to each V2I slice group from the static spectrum resource pool can be obtained as follows:
[0093]
[0094] According to step 1.3, the base station can obtain the QoS indicator transmission rate satisfaction status of each V2I network slice based on the downlink transmission rate of the previous SPS cycle, that is:
[0095]
[0096] If for any V2I slice group s, meetQoS(s)<1 in the previous SPS cycle, it means that the transmission rate of the V2I user QoS indicator in the V2I slice group s is not met, and then RBs are allocated from the dynamic spectrum resource pool to the V2I slice group s at the minimum granularity, i.e., a single time-frequency resource block RB, and this step is repeated until in the current SPS cycle, meetQoS is an all-1 vector, i.e., all three elements in the vector are 1;
[0097] At this time, based on the actual transmission rate of each V2I slice group, if there are still unallocated time-frequency resource blocks RB in the dynamic spectrum resource pool, the average transmission rate of each V2I slice group in its video resource block RB set can be obtained according to step 1.5, that is:
[0098]
[0099] The base station tends to assign it to the V2I slice group that can maximize the μ value in the current SPS period, and Allocate the time-frequency resource block RB to the V2I slice group s. Repeat this process until there are no time-frequency resource blocks RB left in the dynamic spectrum resource pool. This completes the semi-positive definite allocation of the time-frequency resource blocks of the V2I slice group.
[0100] Furthermore, it is assumed that there are 3 V2V NOMA clusters within the coverage of the base station, with a total of 12 V2V Tx users and 20 V2V Rx users. And the 12×3-dimensional V2V Tx and V2V NOMA cluster attribution matrix Affi_NOMA is known. The spectrum sharing mode of the three V2V NOMA clusters and the V2I slice group is known, that is, the time-frequency resource block RB set RB of the three V2V NOMA clusters 1 , RB 2 , RB 3 Assume that the communication distance between V2V users is D Tx =30 meters.
[0101] According to step 2.2-step 2.4, the V2V Rx users within the transmission range of each V2V Tx are classified. For example, when p = 1, if V2V Tx 1 has 1 V2V Rx 10 within its transmission range, and there are no other V2V Tx users in the same V2V NOMA cluster as V2V Tx 1 within the reception range of V2V Rx 10, if V2V Tx 1 belongs to V2V NOMA cluster 2, that is, Affi_NOMA(1,2) = 1. Then set the temporary power variables Temp1_P1, Temp2_P1.
[0102] Step S1: Initialization state, Temp1_P1 = 0, Note | RB 2 | is the number of time-frequency resource blocks (RBs) of V2VNOMA cluster 2, taken as:
[0103]
[0104] Step S2: Optimize the problem P according to step 2.1 of claim 3 * Based on the constraints C2 and C3, calculate the transmission rate R of V2V Rx10 and V2V Tx 1 1,10 ;
[0105] Step S3: Assuming that the broadcast data packet length is 296 bits and the acceptable delay is 0.1 milliseconds, the lower limit of the transmission rate is 2.96 Mb / s. 1,10 ≥2.96Mb / s, then update If R 1,10 <2.96Mb / s, then update
[0106] Step S4: Set error precision ∈=10 -3 , if |Temp1_P1-Temp2_P1|>10 -3 , repeat steps S2 and S3 until |Temp1_P1-Temp2_P1|<10 -3 , and finally the total transmission power of V2V Tx 1 is updated to And take As the transmission power of V2V Tx 1 on RBn.
[0107] For example, if p = 5, if V2V Tx 5 has 1 V2V Rx 12 within its transmission range, and there are other V2V Tx users in the same V2V NOMA cluster as V2V Tx 5 within the reception range of V2V Rx 12, such as V2V Tx 6, and if V2V Tx 5 belongs to V2V NOMA cluster 3, that is, Affi_NOMA(5,3) = 1. Then set the temporary power variables Temp1_P5, Temp2_P5, Temp1_P6, and Temp2_P6.
[0108] Step S5: To initialize:
[0109]
[0110] Step S6: Optimize problem P according to step 2.1 * Based on the constraints C2 and C3, calculate the transmission rate R of V2V Rx 12 and V2V Tx 5 5,12 ;
[0111] Step S7: If R 5,12 ≥2.96Mb / s, then update If R 5,12 <2.96Mb / s, then update
[0112] Step S8: Optimize problem P according to step 2.1 * Based on the constraints C2 and C3, calculate the transmission rate R of V2V Rx 12 and V2V Tx 6 6,12 ;
[0113] Step S9: If R 6,12 ≥2.96Mb / s, then update If R 6,12 <2.96Mb / s, then update
[0114] Step S10: Repeat steps S6-S9 until |Temp1_P5-Temp2_P5|<10 -3 and |Temp1_P6-Temp2_P6|<10 -3 , then determine And take As the transmission power of V2V Tx 1 on RB n; determine And take As the transmission power of V2V Tx 1 on RB n.
[0115] For example, if p = 10, if V2V Tx 10 has two V2V Rx 15 and V2V Rx 16 within its transmission range, and there are other V2V Tx users in the same V2V NOMA cluster as V2V Tx 10, such as V2V Tx11, within the acceptance range of V2V Rx 15, if V2V Tx 10 belongs to V2V NOMA cluster 2, that is, Affi_NOMA(10,2) = 1. There are no other V2V Tx users in the same V2V NOMA cluster as V2V Tx 10 within the acceptance range of V2V Rx 16.
[0116] According to step 2.4.1 of claim 6, two temporary power optimization variables are set, denoted as P 10,1 , P 10,2 They correspond to temporary power optimization variables of V2V Rx 15 and V2V Rx 16 respectively.
[0117] Step S11: For P 10,1 According to steps S5-S9, a temporary power optimization variable P is completed. 10,1 Optimization update, such as updating P 10,1 After that, it makes:
[0118]
[0119] If the value of P is reduced, the update result is retained, otherwise steps S5-S9 are repeated so that the found P 10,1 If the value of the above formula can be reduced, the updated result is retained;
[0120] Step S12: Update according to the least squares principle Update the Lagrange multiplier:
[0121]
[0122] Step S11: For P 10,2 According to steps S1-S3, a temporary power optimization variable P is completed. 10,2 Optimization update, such as updating P 10,2 After that, it makes:
[0123]
[0124] If it decreases, keep the updated result, otherwise repeat steps S1-S3 so that the found P 10,2 If the value of the above formula can be reduced, the updated result is retained;
[0125] Step S12: Update according to the least squares principle Update the Lagrange multiplier:
[0126]
[0127] Step S15: Repeat steps S11-S14 until P 10,1 , P 10,2 The updated reading meets the requirements and finally obtains The power allocation result.
Claims
1. A semi-positive definite resource allocation method for Internet of Vehicles based on network slicing and NOMA, It is characterized in that The following steps are involved: Step 1: Vehicle users communicating with infrastructure are referred to as V2I users, and V2I user groups in the same network slice are referred to as V2I slice groups. When the number of V2I slice groups and the transmission rate of the user communication quality QoS indicator in each V2I slice group are known, the static spectrum resource pool is used to initialize the time-frequency resource block RB allocation for each V2I slice group. Furthermore, a semi-positive definite resource allocation time period is referred to as an SPS period, and an SPS period includes a resource reservation management phase and a data transmission phase. A minimum granularity of allocable time-frequency resources is referred to as a time-frequency resource block RB. The base station uses a dynamic spectrum resource pool to fine-tune the time-frequency resource block RB allocation of each V2I slice group according to the downlink data transmission rate fed back by each V2I slice group in the data transmission phase of the previous SPS period, until the QoS indicator transmission rate of each V2I slice group is met. If there are still excess time-frequency resource blocks RB that have not been allocated, the remaining sub-channels are allocated one by one with different probabilities according to the data transmission rate of each V2I slice group until they are fully allocated. Step 2: The group of transmitting users of vehicle-to-vehicle communication that are classified into the same non-orthogonal multiple access cluster is recorded as a V2V NOMA cluster, and the vehicle users of vehicle-to-vehicle communication are recorded as V2V users; the transmitting users of vehicle-to-vehicle communication are recorded as V2V Tx users; and the receiving users of vehicle-to-vehicle communication are recorded as V2V Rx users. In the data transmission phase of the current SPS cycle, assuming that the spectrum sharing mode of the V2V NOMA cluster and each V2I slice group and the V2V Tx users in each V2V NOMA cluster are known, it is stipulated that V2V Rx receives the broadcast signal of V2V Tx within a specific receiving range; V2V Tx users and V2V Rx users within the receiving range perform distributed transmission power control updates on V2V Tx users through an iterative algorithm, and improve the transmission energy efficiency of V2V users while meeting the delay requirements. in: The step 1 is specifically as follows: Step 1.1: Assume that the transmission power of the base station on each time-frequency resource block RB is fixed, and assume that there are M V2I users within the coverage area of the base station. Define the counting variable m, 1≤m≤M, and record the mth V2I user as V2I m. The QoS indicator transmission rate of each V2I user is recorded as The number of V2I slice groups is recorded as S, and a counting variable s is defined, 1≤s≤S, and an M×S-dimensional V2I user and V2I slice group belonging discrimination matrix Affi_NS is defined, where Affi_NS(m, s)=1 indicates that V2I m belongs to V2I slice group s, otherwise Affi_NS(m, s)=0; Define an S-dimensional vector aveQoSrate, where the sth element represents the average QoS indicator transmission rate of V2I slice group s, that is: Step 1.2: The allocatable time-frequency resources in the base station are divided into a static time-frequency resource pool and a dynamic time-frequency resource pool. According to the vector aveQoSrate in step 1.1, the time-frequency resource block RB ratio of the static spectrum resource pool of each V2I slice group is determined. The static time-frequency resource block ratio of V2I slice group s is recorded as The total bandwidth of the time-frequency resource blocks in the static spectrum resource pool is Then the bandwidth allocated by V2I slice group s from the static spectrum resource pool in the initialization phase is Step 1.3: The actual transmission power of V2I m is Each V2I user uploads the downlink transmission rate of the previous SPS cycle through the SPS cycle uplink The base station collects statistics on the QoS indicator transmission rate satisfaction of each V2I slice group, defines an S-dimensional QoS indicator transmission rate satisfaction discriminant vector meetQoS, initializes it to all zeros, and the sth element can represent the satisfaction of the QoS indicator transmission rate of the V2I slice group s, that is: Where sgn(x) is a sign function, when x>0, sgn(x)=1; when x=0, sgn(x)=0; when x<0, sgn(x=-1; if meetQoS(s)=1, the QoS indicator transmission rate of all V2I users in the V2I slice group s is met, otherwise the QoS indicator transmission rate of some V2I users in the V2I slice group s is not met; Step 1.4: If for any V2I slice group s, meetQoS(s) in the previous SPS cycle is less than 1, allocate RBs from the dynamic spectrum resource pool to the V2I slice group s at the minimum granularity, i.e., a single time-frequency resource block RB, and repeat step 1.4 until meetQoS is a vector of all 1s in the current SPS cycle; Step 1.5, set the S-dimensional dynamic allocation variable μ. If after step 1.4, meetQoS is a vector of all 1s, and there are still idle time-frequency resource blocks RB in the dynamic spectrum resource pool, then for each idle time-frequency resource block RB, in order to properly consider fairness, it tends to be allocated to the V2I slice group that can maximize the μ value in the current SPS period, with probability The time-frequency resource blocks RB are allocated to the V2I slice group s, where: Where |RB s | represents the number of time-frequency resource blocks (RBs) of V2I slice group s, RB s represents the time-frequency resource block RB set of the V2I slice group s; repeat step 1.5 until there are no idle time-frequency resource blocks RB in the dynamic spectrum resource pool; The step 2 is specifically as follows: Step 2.1: Assume that there are J V2V NOMA clusters, P V2V Tx users, and Q V2VRx users within the coverage area of the base station. Define a counting variable p. Define a P×J-dimensional matrix Affi_NOMA, Affi_NOMA(p, j) = 1 indicates that V2VTx p belongs to V2V NOMA cluster j; otherwise, Affi_NOMA(p, j) = 0; the set of time-frequency resource blocks in V2V NOMA cluster j is RB j , the following optimization problem is constructed for the power control of V2V Tx in a single V2V NOMA cluster. Taking V2V NOMA cluster j as an example, the optimization problem P * The build is as follows: Optimization Problem P * The five constraints C1 to C5 are included. The pth V2V Tx user is denoted as V2V Tx p; the qth V2VRx user is denoted as V2V Rx q; a counting variable n is defined, and the nth time-frequency resource block RB is denoted as RB n. represents the transmission power of V2VTx p on RB n, is the channel gain of V2V Tx p and V2V Rx q on RB n, is the channel gain from the base station to V2V Rx q on RB n, The channel gain on RB n is weaker than The channel gain between V2V Tx p′ and V2V Rx q; is the downlink power of the V2I user that is the nearest neighbor to V2V Rx q and shares spectrum resources with V2V NOMA cluster j; is the maximum transmission power of each V2V Tx user, is the transmission rate of V2V Tx p and V2V Rx q on RB n, R pq is the transmission rate between V2V Tx p and V2V Rx q, L is the broadcast packet length, D is the maximum tolerable delay, B RB is the bandwidth of a single RB; σ 2 is the noise power; Step 2.2: The communication distance between V2V users is recorded as D Tx , then in the V2V Tx p radius D Tx All V2V Rx users within the range can receive the broadcast data packets of V2V Tx p. Therefore, only the V2V Rx users and V2V Tx p that can receive the V2V Tx p signal are considered to jointly optimize the power control of V2V Tx p. Tx There is only one V2V Rx q within the radius D of the V2V Rx q. Tx If there is no other V2V Tx user in the same V2V NOMA cluster as V2V Tx p within the range, set the temporary power variable Temp1_P p , Temp2_P p ; Step 2.3: If the radius D of V2V Tx p Tx There is only one V2V Rx q within the radius D of the V2V Rx q. Tx If there are other V2V Tx users in the same V2I slice group as V2V Tx p within the range, a rotation strategy is adopted to optimize the radius D of the V2V Rx q. Tx Power control of each V2V Tx p within; Step 2.4: If the radius D of V2V Tx p Tx Inside, there is L p V2V Rx, define the counting variable l, then this L p The V2V Rx is denoted as q l Through this L p Each V2V Rx performs joint optimization on the transmission power of the V2V Tx p; The step 2.2 is specifically as follows: Step 2.2.1, Initialization state, Temp1_P p =0, Note | RB j | is the number of time-frequency resource blocks (RBs) of V2V NOMA cluster j, and is taken as: Step 2.2.2: According to the optimization problem P * Based on the constraints C2 and C3, calculate the transmission rate R of V2V Rx q and V2V Tx p pq ; Step 2.2.3: If Update like Update Step 2.2.4, set the error precision ∈, if |Temp1_P p -Temp2_P p |>∈, repeat steps 2.2.1 to 2.2.3 until |Temp1_P p -Temp2_P p |<∈, the total transmission power of V2V Tx p is finally updated as And take as the transmission power of V2V Tx p on RB n; The step 2.3 is specifically as follows: Step 2.3.
1. Assume that the radius of V2V Rx q is D Tx There is K j The V2V Tx users of a V2V NOMA cluster j have j The channel gains between V2V Tx users and V2V Rx q are arranged in descending order, that is, For this K j V2V Tx user settings 2K j A temporary power variable, denoted as Where k is a counting variable, Step 2.3.2: initialization Pick: Step 2.3.3, k = 1, according to the constraints C2 and C3, without changing other K j -1 V2V Tx user transmission power, calculate the transmission rate R of V2V Rx q and V2V Tx p pq ; Step 2.3.4: If Update like Update renew As the transmission power of V2V Tx p on RB n, the total transmission power of V2V Tx p is Step 2.3.5, k = k + 1, repeat step 2.3.3 ～ Step 2.3.4, repeat this step until K j V2V Tx users are traversed; Step 2.3.6, replace step 4.4.2 ～ Step 4.4.5 is a cycle, and it is repeated multiple times until End the loop; The step 2.4 is specifically as follows: Step 2.4.
1. Define L for V2V Tx p p Temporary power optimization variables, denoted as initialization Set the iteration round count variable t and initialize t=0; Furthermore, for the optimization problem P * The objective function is modified to: And add new constraints: Construct the augmented Lagrangian function: where v l is the Lagrange multiplier, initialized to v l (t), is the penalty term; the above formula is equivalent to: Step 2.4.2: When l = 1, through P pl With V2V Rx q l The joint power optimization of V2V Rx q l The radius D Tx If there are no other V2V Tx users in the same V2V NOMA cluster as V2V Tx p within the range, implement step 2.2.1 ～ Step 2.2.4, as updated After If it decreases, keep the updated result, otherwise repeat step 2.2.1 ～ Step 2.2.4, find the Reduced and update Step 2.4.3, if in the V2V Rx q l The radius D Tx If there are other V2VTx users in the same V2I slice group as V2V Tx p within the range, implement step 2.3.1 ～ Step 2.3.5, as updated After If it decreases, keep the updated result, otherwise repeat step 2.3.1 ～ Step 2.3.5, find the Reduced and update Step 2.4.4, Update Make According to the principle of least squares, we know Step 2.4.5, update the Lagrange multiplier: Step 2.4.6, l = l + 1, repeat step 2.4.1 ～ Step 2.4.5, until L temporary power optimization variables are traversed; Step 2.4.7: Replace step 2.4.1 ～ Step 2.4.6 is repeated as one round, defining concise reading ∈ 2 , until L equal The decrease is less than ∈ 2 , stop updating, and take The power allocation result.