CN109905864B - A cross-layer resource allocation scheme for power Internet of things - Google Patents

Abstract

The invention mainly relates to a cross-layer resource allocation scheme applied in the power Internet of things, and realizes long-term stability of the queue by optimizing the data queue transmitted to the base station by multiplexing the channels of other user equipment in the cellular network by various machine communication equipment. By studying Lyapunov optimization and Gal-Shalip matching algorithm, a cross-layer rate control and resource allocation mechanism is proposed. The algorithm proposed by the present invention mainly converts the original long-term optimization problem into a rate control sub-problem and a resource allocation sub-problem in each time slot. First, the Lyapunov algorithm is used to decompose two convex functions, and the algorithm complexity can be used. The lower convex optimization tools are easy to solve. For the sub-channel selection problem, the bidirectional preference list of machine pairs and cellular UEs is first established according to the different transmission performance of different sub-channels, and then the iterative Gal-Shalip algorithm is used. Complete the final stable match. Simulation results show that the present invention can significantly improve queue stability and optimize network performance without prior knowledge of data arrival and sub-channel statistics.

Description

A cross-layer resource allocation scheme for power Internet of things

technical field

The invention belongs to the field of wireless communication, and in particular relates to a cross-layer resource allocation scheme for machine-to-machine (M2M) communication applied in the power Internet of things. . First, the randomly arriving data queues are optimized by Lyapunov optimization algorithm, and then the Gale-Shalip matching theory is used to optimize the data queues with different priorities in the network using Orthogonal Frequency Division Multiplexing (OFDM). Suitable for sub-channel selection, thereby maximizing resource utilization and minimizing network delay.

Background technique:

With the rapid development of communication technology and the large-scale access of data collection terminals, the era is experiencing a huge transition from traditional human-to-human communication to machine-to-machine communication. As one of the key supporting technologies for the networking and operation of the Internet of Things, M2M communication is crucial for the implementation of industrial automation and smart grids because of its excellent self-organization and self-healing capabilities. As far as the smart grid is concerned, the research and construction of the power Internet of things has been in full swing, and the existing mobile cellular network also provides a good foundation for its popularization, but although the M2M communication technology in the power Internet of things has been widely studied and applied, However, there are still some urgent problems and challenges that need to be solved, which are summarized as follows:

1) Stability of transmission queues: In real scenarios, because the arrival of data streams is dynamic and unpredictable, coupled with the influence of time-varying channels, data transmission queues are often not distributed smoothly, which brings difficulties to the operation of base stations. A lot of pressure. In order to solve this problem, avoid channel congestion and reduce the packet loss rate, an effective device access control scheme is very necessary.

2) Performance optimization for user experience: The rapid growth of data and traffic will inevitably lead to insufficient wireless spectrum resources, which is also an important factor limiting the quality of user experience. Unfortunately, current research usually ignores the subjective feelings of users and focuses more on optimizing network performance only. Therefore, how to utilize limited spectrum resources to improve user experience quality is an important challenge currently.

3) Long-term system performance optimization: In the power Internet of Things, massive real-time data generated by various data acquisition terminals with different functions floods into the existing cellular network, which will overwhelm the base station, and even the control system composed of these devices. The system crashes, and the current mainstream network performance optimization algorithms (such as queuing theory) usually only achieve short-term optimization. However, the complex environment always brings a lot of uncertainty and randomness to the data transmission process. Therefore, considering the above-mentioned dynamic influencing factors, it is an urgent problem to design a long-term performance optimization scheme to make the data queue stable and resource utilization higher.

Based on the problems and challenges mentioned above, the present invention mainly proposes a cross-layer resource allocation scheme for the power Internet of Things, in which the Lyapunov optimization and the Gal-Shalip algorithm are jointly applied to the OFDM cellular network with spectrum sharing M2M communication to maximize network performance and meet the needs of dense users.

Invention content:

The invention firstly simulates the coexistence scenario of multiple ordinary cellular users in one cell and machine communication pairs in the power Internet of Things, aiming at smooth distribution of data transmission queues and maximization of resource utilization, and proposes a power Internet of Things-oriented cross-layer resource allocation scheme. This solution considers the experience requirements of the cellular user equipment and the M2M pair, first completes the sub-channel selection of the latter multiplexing the former according to the current known information, sends the collected data to the base station, and optimizes the received data queue on the base station side processing, and control the arrival of the next data and the selection of sub-channels according to the optimization situation, so as to quickly solve the problems of base station overload and user experience quality degradation. The specific process is as follows:

1) Establishment of the queue model

Fig. 1 is a cellular network system model based on M2M uplink, which consists of a base station, N M2M pairs, and K cellular user equipments. The base station is responsible for resource coordination and sub-channel allocation in the cell. Different cellular user equipments will correspondingly generate K mutually orthogonal and non-interfering sub-channels. The M2M pair is composed of a transmitter (MT) and a receiver (MR) respectively. The composition, in our scenario, only considers the uplink part where the transmitter transmits data to the base station.

和

对应产生K个正交子信道，表示为

假设时隙t内的M_n的数据准入速率表示为A_n(t)，对应传输率为R_n(t)，当前数据积压队列为Q_n(t)。数据准入速率A_n(t)是Q_n(t)的输入，Q_n(t)是网络层参数，传输速率R_n(t)则是输出，它是物理层参数。Q_n(t)随着时间的变化如下：In the system, the discrete time model is adopted, and every 1 second is regarded as a time slot t, in the total duration of T. Within the communication range of the base station, the number of machine communication pairs and cellular user equipment remains the same, but their positions are randomly distributed in different time slots for randomness considerations. In the time slot t, it is assumed that there are M vehicles and K user equipments, respectively expressed as

and

Correspondingly, K orthogonal sub-channels are generated, which are expressed as

Assuming that the data admission rate of _Mn in time slot t is represented as An (t), the corresponding transmission rate _{is R n (} t), and the current data backlog queue is Q _n (t). The data admission rate An( _t ) is the input of Qn( _t ), Qn(t) is the network layer parameter, and the transmission rate _Rn ( _t ) is the output, which is the physical layer parameter. Q _n (t) varies with time as follows:

Q _n (t+1)=[Q _n (t)-R _n (t)] ⁺ +A _n (t)

where [x] ⁺ means max(x, 0). And we consider Q(t) to be strongly stable when it satisfies the following conditions:

To achieve dynamic queue stabilization, we need to control A _n (t) and R _n (t) separately, which will be introduced later.

2) Establishment of MOS (Mean Opinion Score) evaluation model

An ( _t ) is a parameter that reflects the performance of the network layer, which directly affects the user experience quality. In some cases, the uplink network load is so heavy that it cannot meet the quality requirements of all users under poor channel conditions, which means that the corresponding data rate needs to be adjusted according to the user experience quality to avoid congestion. Incoming rate adjustment is also known as rate control. In order to characterize the quality of user experience with mathematical methods, we established the following MOS evaluation model, as follows:

MOS[A _n (t)]=η _n log ₂ [A _n (t)]

Among them, A _n (t) represents the data admission rate of _Mn at time t, and the parameter η _n ∈ [0, 1] represents the priority parameter set for Mn, and the larger η _n is, the greater the value of η _n , the greater the delay of the data generated by _Mn . The higher the requirements, the more communication resources should be occupied.

3) Transmission channel modeling

Uplink communication resource allocation (eg, power optimization and sub-channel selection) occurs at the beginning of each time slot. In an OFDM system in which cellular user equipment and M2M pairs coexist, the bandwidth is equally divided into K sub-channels, and the bandwidth of each sub-channel is B.

The Signal to Interference plus Noise Ratio (SINR) channel SNR of C _k sharing the subchannel with _Mn can be expressed as:

表示C_k与基站之间的距离，α_C表示所有蜂窝用户设备在当前场景下的路径损耗参数。N₀表示环境的加性高斯白噪声大小。对应的，p_n(t)、g_nk(t)、

和α_M分别表示时隙t内MT_n的发射功率、发射功率增益、MT_n与基站之间的距离以及路径损耗参数。where p _k (t) represents the transmit power of C _k in time slot t, g _k (t) represents the transmit power gain of C _k in time slot t,

represents the distance between C _k and the base station, and α _C represents the path loss parameter of all cellular user equipments in the current scenario. N ₀ represents the additive white Gaussian noise level of the environment. Correspondingly, p _n (t), g _nk (t),

and α _M represent the transmit power, transmit power gain, distance between MT _n and the base station, and path loss parameters of MT _n in time slot t, respectively.

Similarly, the SINR backhaul channel SNR from the base station to the M2M pair receiver is:

表示MT_n与MR_n之间的距离，

表示C_k与MR_n之间的距离。则在时隙t内，M_n的MT和MR之间复用信道S_k形成的的传输速率为：in,

represents the distance between MT _n and MR _n ,

represents the distance between C _k and MR _n . Then in time slot t, the transmission rate formed by multiplexing channel _Sk between MT and MR of _Mn is:

表示一个关于是否在时隙t将C_k占用的S_k分配给M_n的二值决策变量。

意味着M_n复用信道S_k。in,

represents a binary decision variable about whether to assign _Sk occupied by _Ck to _Mn at slot t.

means that _Mn multiplexes the channel _Sk .

在每个时隙t只能由至多一个M2M对重用，以避免对C_k和BS之间的现有上行链路的过度干扰。因此，我们有each subchannel

It can only be reused by at most one M2M pair per slot t to avoid excessive interference to the existing uplink between _Ck and the BS. Therefore, we have

4) Long-term data admission rate constraints and delay constraints

Since there are many delay-sensitive devices, which usually require a delay upper bound and a rate transfer rate lower bound, we impose a time-averaged rate constraint and a delay constraint on each M2M pair.

Specifically, the time-averaged data admission rate constraints are as follows:

_where On represents the minimum long-term data admission rate for _Mn .

Queuing delay is generally defined as the length of time a packet waits in a queue until it can be transmitted. It should be noted that the transmission delay in a network with high load is small compared to the queuing delay and thus can be ignored. The formula for the average delay constraint defined in time is as follows:

Among them, ρ _n represents the average delay, and its upper bound is D _n .

5) Modeling of the maximum MOS optimization problem

用于表示子信道选择策略，P＝{p_n}用于表示功率优化策略，R＝{R_n}用于表示数据准入速率控制策略。优化问题被建模如下：The optimization of the weighted MOS for all M2M pairs requires solving joint rate control, power optimization and subchannel selection problems and involves two-dimensional matching between M2M pairs and subchannels. So at time slot t, let a two-dimensional matrix of size N × K

is used to represent the sub-channel selection strategy, P={p _n } is used to represent the power optimization strategy, and R={R _n } is used to represent the data admission rate control strategy. The optimization problem is modeled as follows:

C ₆ : Queue Q _n (t) is strongly stable,

Among them, the constraints C1 and C2 are to ensure that each subchannel can be reused by at most one M2M pair in each slot, and vice versa. C3 specifies the transmit power constraints for the M2M pair. C4 is the SINR threshold constraint for a cellular user equipment and M2M pair. C5 is the maximum bearable data queue rate of the base station. C6 is the stability constraint of the M2M pair. C7 and C8 ensure that the rate requirement and time-averaged delay of each subchannel of the M2M pair are guaranteed at the same time.

6) Problem solution based on Lyapunov optimization algorithm

To simulate average delay and rate constraints, we introduce the concept of virtual queues. The virtual queue Y(t) associated with the average rate constraint varies over time as follows:

Y _n (t+1)=[Y _n (t)-A _n (t)] ⁺ ₊ On (t)

If the virtual power queue Y(t) is average rate stable, then it satisfies the average power constraint _C7 .

The virtual queue Z(t) associated with the delay constraint varies over time as follows:

Z _n (t+1)=[Z _n (t)-D _n R _n (t)] ⁺ +Q _n (t)

According to the above analysis, if the data queue and the two virtual queues (Y, Z) are stable for all M2M pairs, we consider the entire network to be stable, and the long-term data admission rate constraints and delay constraints are satisfied .

Therefore, we can transform the original optimization problem of step 6) into a problem of maximizing the weighted MOS value of all M2M pairs according to the queue stability constraints C ₁ , C ₂ , C ₃ , C ₄ and C ₅ . The transformed problem is formulated as follows:

stC ₁ , C ₂ , C ₃ , C ₄ and C ₅

_C6 : Queues Q(t), _Yn ( _t ) and Zn(t) are strongly stable,

Let Q={Qn(t)}, Y={ _Yn ( _t )} and Z= _{ Zn(t)} denote the backlog of three queues respectively, let G(t)=[Q(t), Y(t), Z(t)] represents the combined queue backlog of M2M pairs, then the Lyapunov equation can be defined as:

Then the instantaneous (from one time slot to the next time slot) Lyapunov drift Δ(G(t)) is defined as:

Subtracting the average expectation of the weighted MOS value from it, we get the drift negative return term as follows:

where V is a non-negative tunable parameter. According to the design principle of Lyapunov optimization, appropriate rate control and resource allocation decisions should be selected to minimize the upper limit of the drift negative reward term for each time slot t, namely:

where X is a non-negative constant that satisfies the following inequality in all time slots t:

We transform the original optimization problem into an upper bound that minimizes the negative return of drift, again at each slot t, this formula is also subject to resource allocation constraints C ₁ , C ₂ , C ₃ and C ₄ and rate control constraint C ₅ effects. Therefore, the original stochastic network long-term optimization problem is transformed into a series of continuous transient static optimization sub-problems, which can be divided into data admission rate control sub-problems and resource allocation sub-problems.

Data admission rate control: The admission rate control strategy means that the algorithm adjusts the admission rate related to the MOS based on the requirements of the M2M pair and the current data queue backlog. For example, in the presence of a backlog of data queues, an M2M pair will reject with a very high probability newly arriving data with a higher priority and a larger amount to avoid more severe channel congestion. In addition, for the case of a certain fixed sub-channel, assuming that the non-negative adjustable parameter V is larger, the M2M pair can adopt a more relaxed admission rate control strategy to allow more data to be accepted. Therefore, the corresponding MOS value of the M2M pair can reach a higher level.

Since the second term of the drift negative reward only involves the admission rate control related parameter An ( _t ), the minimization of this term can be regarded as the first sub-problem, as follows:

stC ₅

因为MOS[A_n(t)]是关于A_n(t)的一个凸函数，我们可以直接应用凸函数优化工具对上式进行优化。in,

Because MOS[A _n (t)] _is a convex function of An (t), we can directly apply the convex function optimization tool to optimize the above formula.

因此对该项的最小化可视为第二个子问题，具体如下：Resource allocation sub-problem: Since the third term of the drift negative return only involves the resource allocation related parameters R _n (t), that is, the power allocation result p _n (t) and the sub-channel selection result

Therefore, the minimization of this term can be regarded as the second sub-problem as follows:

stC ₁ , C ₂ , C ₃ and C ₄

资源分配问题是一个较为复杂的组合问题，其中变量

是离散的，但变量p_n(t)是连续的。在实际应用中，由于其高复杂性，难以实现对最优值的详尽搜索，但通过应用逐次超松弛的方法，原始整数规划问题可以放宽到凸优化问题，该方法复杂度较低，易于找到满足约束条件的最佳解决方案。in,

The resource allocation problem is a more complex combinatorial problem, in which the variable

is discrete, but the variable p _n (t) is continuous. In practical applications, it is difficult to achieve an exhaustive search for the optimal value due to its high complexity, but by applying the method of successive over-relaxation, the original integer programming problem can be relaxed to a convex optimization problem, which has low complexity and is easy to find The best solution that satisfies the constraints.

表示从集合

到其自身的一个一对一映射关系，φ(M_n)＝S_k表示M_n与子信道S_k匹配，此时

否则为0。In addition, we solve the sub-channel selection problem by applying the Gale-Shalip matching algorithm, that is, a two-dimensional matching problem involving N M2M pairs and K sub-channels. First we give the following definition: match

represents from the collection

One-to-one mapping relationship to itself, φ(M _n )=S _k indicates that M _n matches the sub-channel _Sk , at this time

0 otherwise.

的最优值可以通过下列关系被求出：When φ(M _n )=S _k , in other words, when

The optimal value of can be found by the following relationship:

stC ₁ , C ₂ , C ₃ and C ₄

This is also a convex optimization problem, and the optimal solution can be directly solved by the convex optimization tool.

6) Problem solution based on Gale-Shalip matching algorithm

表示，通过将每个M2M对与每个子信道短暂连通以获得与每个子信道对应的传输速率，速率越高，优先级越高。每个蜂窝用户设备是否愿意将子信道提供给M2M对则由SINR值的大小决定，对用户本身造成的干扰越大，即SINR值越小，则越不愿意。进而通过迭代的匹配算法最终实现一个稳定的匹配φ。其基本流程如下：In order to complete the sub-channel selection, we first need to establish the bidirectional preference list of the M2M pair and the cellular user equipment. We define each M2M's preference for different subchannels by its transmission rate

means that the transmission rate corresponding to each subchannel is obtained by briefly connecting each M2M pair with each subchannel, and the higher the rate, the higher the priority. Whether each cellular user equipment is willing to provide subchannels to the M2M pair is determined by the size of the SINR value. Then, a stable matching φ is finally achieved through an iterative matching algorithm. The basic process is as follows:

Step 1: Each _Mn makes a matching application to _Sk , which ranks first in its preference list and has not expressed rejection to itself;

Step 2: If the corresponding _Sk has not been selected, the two are successfully matched. If it has been selected, compare the order of the applicant and the original matcher in the sub-channel to, select the first M2M pair, and reject it at the same time. another;

Step 3: Repeat steps 1 and 2 until each _Mn selects one of the sub-channels and is rejected by all the remaining sub-channels.

Description of drawings:

Figure 1 is a cellular network system model based on M2M uplink.

FIG. 2 is the simulation parameters when the present invention performs simulation.

FIG. 3 is the result of the queue control proposed by the present invention.

Fig. 4 is the resource allocation result proposed by the present invention.

FIG. 5 is a comparison between the system performance proposed by the present invention and the result stability of the random matching algorithm.

Detailed ways

The implementation of the present invention is divided into two steps, the first step is to establish a model, and the second step is to implement the algorithm. The established model is shown in FIG. 1 , which completely corresponds to the introduction of the M2M uplink-based cellular network system model in the Summary of the Invention.

1) For the system model, with the extensive construction of the Internet of Things in electric power, a large amount of data floods into the existing mobile cellular network, but due to the dynamic and unpredictable arrival of the data flow, it brings great pressure to the operation of the base station. There is an urgent need to design a long-term stable optimization scheme for a large amount of data when the global information is unknown. As shown in Figure 1, the M2M pair in the power Internet of Things transmits data by multiplexing the channel of the appropriate cellular user equipment, and applies the optimization method of the random network to control the data arriving at the base station to achieve long-term stability of the transmission queue. Greatly improve network performance and user experience quality.

2) In order to solve the above optimization problem, a Lyapunov-based data admission rate control scheme should be designed first, and the Gale-Shalip matching algorithm should be combined to achieve a reasonable allocation of communication resources. Finally, the design is decomposed into two sub-problems of convex function optimization of data admission rate control and resource allocation, which can be well solved by the convex optimization toolbox with lower complexity, supplemented by Gal-Shalip. Matching algorithms to achieve sub-channel selection.

For the present invention, we performed extensive simulations. The specific parameters in the simulation are shown in Table 2. 4 M2M pairs and 5 cellular user equipments are randomly distributed in the cellular network with a radius of R=200m. The results are described in the following two aspects: data queue rate control and power delay.

Figure 3 shows the backlog changes for different queues and time slots. We can observe that each queue tends to settle around the corresponding value after only a few slots when given a random initial backlog. Numerical results demonstrate that rate control can be well achieved by utilizing our method to deal with the continuously generated backlog queue. It is worth mentioning that the size of the backlog is positively related to the priority, because the M2M pair with higher priority has more data collection volume and more frequent data sending, resulting in more queue backlog.

Figure 4 shows the joint power optimization and sub-channel selection scheme for different time slots. In particular, similar stable results for virtual queue Z are given in Fig. 3-(a), where this value represents the total power to be allocated _Pmax . Figure 3-(b) shows the power optimization results, where the four gradients differ for different priorities of the M2M pair. After allocation, the transmission rate of the sub-channel is shown in Figure 3-(c). The above results show that the combined algorithm of Lyapunov optimization and Gale-Shalip matching can not only maintain the stability of the system, but also avoid the adverse effects of time-varying channels as much as possible.

Figure 5 compares the overall system stability of our proposed algorithm and Lyapunov optimization combined with random matching from the perspective of queue backlog and power allocation, where two box plots are shown to show the dispersion of a set of data . It can be seen from the figure that whether it is queue backlog or power allocation, the overall distribution of the proposed scheme is more concentrated than that of random sub-channel selection, which is because random matching exists to compare poor-performing sub-channels with high-priority sub-channels. The possibility of a high queue match.

Although the specific implementation of the present invention and the accompanying drawings are disclosed for no stated purpose, the purpose is to help understand the content of the present invention and implement it accordingly, but those skilled in the art can understand that: without departing from the present invention and the appended claims Various substitutions, changes and modifications are possible within the spirit and scope. Therefore, the present invention should not be limited to the contents disclosed in the best embodiments and the accompanying drawings, and the scope of protection of the present invention shall be subject to the scope defined by the claims.

Claims (1)

1. a cross-layer resource allocation method oriented to the power Internet of things, is characterized in that: 1) Consider using the Lyapunov optimization method to optimize the data queues transmitted to the cellular by various machine-type communication devices M2M in the power Internet of Things when the relevant data queue and channel statistics are unknown, and convert the long-term optimization problem into a series of Rate control sub-problem and resource allocation sub-problem within time slot t to finally achieve long-term stationary state and optimize network performance; specifically include: (1) First, establish a queue model. The change of the queue _backlog corresponding to M2M to MN with time slot t is as follows: Q _n (t+1)=[Q _n (t)-R _n (t)] ⁺ +A _n (t) [x] ⁺ represents max(x, 0), R _n (t) represents the transmission rate, An ( _t ) represents the data admission rate, (2) Set the MOS evaluation model as: MOS[A _n (t)]=η _n log ₂ [A _n (t)] The parameter η _n ∈ [0, 1] represents the priority parameter set for _Mn , (3) In an OFDM system where cellular user equipment and M2M pairs coexist, the bandwidth is equally divided into K sub-channels, the bandwidth of each sub-channel is B, and the transmission channel is modeled as follows:

represents a binary decision variable on whether to assign the subchannel _Sk occupied by the cellular user equipment C _k to _Mn at time slot t,

represents the SINR backhaul channel signal-to-noise ratio from the base station to the M2M pair receiver at this time, and the previous uplink transmission signal-to-noise ratio SINR value is expressed as

The correspondences are obtained by the following equations, respectively:

p _k (t) represents the transmit power of C _k in time slot t, g _k (t) represents the transmit power gain of C _k in time slot t,

represents the distance between C _k and the base station, α _c represents the path loss parameter of all cellular user equipments in the current scenario, N ₀ represents the additive white Gaussian noise of the environment, correspondingly, p _n (t), g _nk ( t),

and α _M represent the transmit power of MT _n , the transmit power gain, the distance between MT _n and the base station, the distance between MT _n and MR _n , the distance between C _k and MR _n , and the path loss in time slot t, respectively parameter, (4) Long-term data admission rate constraints and delay constraints are specified as follows:

(5) Finally, the maximum weighted MOS optimization problem is established as follows:

_stC1 :

_C2 :

_C3 :

_C4 :

_C5 :

C ₆ : Queue Q _n (t) is strongly stable,

_C7 :

_C8 :

It is used to represent the sub-channel selection strategy, P={p _n } is used to represent the power optimization strategy, R={R _n } is used to represent the data admission rate control strategy, and constraints C1 and C2 are used to ensure that each sub-channel is used in each A timeslot can be reused by at most one M2M pair and vice versa, C3 specifies the transmit power constraint for the M2M pair, C4 is the SINR threshold constraint for the cellular UE and M2M pair, C5 is the base station's maximum bearable data queue rate, and C6 is the M2M The stability constraints of the pair, C7 and C8 ensure that the rate requirement and time-averaged delay of each subchannel of the M2M pair are guaranteed at the same time, (6) In order to solve the above optimization problem, a virtual queue related to the average rate and power delay is introduced: Y _n (t+1)=[Y _n (t)-A _n (t)] ⁺ ₊ On (t) Z _n (t+1)=[Z _n (t)-D _n R _n (t)] ⁺ +Q _n (t) Then the above problem is transformed into:

stC ₁ , C ₂ , C ₃ , C ₄ and C _{5 C6} : Queues Q(t), _Yn ( _t ) and Zn(t) are strongly stable,

The combined queue backlog of the three queues of the M2M pair is:

Then the instantaneous Lyapunov drift Δ(G(t)) is:

X is a non-negative constant,

Instantaneous is from one time slot to the next time slot; The rate control subproblem is formulated as:

stC ₅ in,

It can be solved using low-complexity convex optimization tools,

stC ₁ , C ₂ , C ₃ and C ₄ in,

exist

In the case of certainty, the convex optimization tool can also be used to solve; 2) In the optimization process, the Gal-Shalip matching algorithm is applied to complete the sub-channel selection, that is, the M2M pair completes data transmission by reusing the channels of the original users in the cellular network, including: First, a bidirectional preference list of M2M pairs and cellular user equipments needs to be established. The preference of each M2M for different subchannels is determined by its transmission rate.

Represents that the transmission rate corresponding to each sub-channel is obtained by connecting each M2M pair to each sub-channel briefly; whether each cellular user equipment is willing to provide the sub-channel to the M2M pair is determined by the size of the SINR value. The greater the interference caused, that is, the smaller the SINR value, the less willing it is; Then a stable matching φ is finally achieved through an iterative matching algorithm; the specific process includes: Step 1: Each _Mn makes a matching application to _Sk , which ranks first in its preference list and has not expressed rejection to itself; Step 2: If the corresponding _Sk has not been selected, the two are successfully matched. If it has been selected, compare the order of the applicant and the original matcher in the sub-channel, select the first M2M pair, and reject the other. one; Step 3: Repeat steps 1 and 2 until each _Mn selects one of the sub-channels and is rejected by all the remaining sub-channels.

Images