Scheduling control method of Internet of things data uplink based on TS algorithm

Abstract

In order to ensure the efficiency of link scheduling control and improve the accuracy and energy efficiency of scheduling control, a scheduling control method for Internet of things data uplink based on TS algorithm is proposed.The principle of tabu search algorithm is analyzed, and the scheduling control function of Internet of things data uplink is constructed. Tabu search algorithm is used to solve the scheduling control problem of Internet of things data uplink, and realize the scheduling control of Internet of things data uplink.The experimental results show that the scheduling control energy efficiency of the proposed method is as high as 10.4 bit/j, the link scheduling control packet loss rate of the proposed method is only 1.4%, and the scheduling control energy efficiency is as high as 10.4 bit/j, which can effectively improve the scheduling control accuracy and energy efficiency.

Keywords

Tabu search algorithm Internet of things data uplink link scheduling control objective function

1. Introduction

With the rapid development of network technology and communication technology, it has played a great role in the development of national economy. One of the key technologies of the Internet of things is a time sensitive physical system, and the Internet of things is a key link in the integration of industrialization and industrialization. It will promote the rapid combination of manufacturing industry and new generation technology [1,2,5]. The purpose of the Internet of things is to efficiently integrate and use massive information to ensure the deterministic behavior of the network physical system under control. A real-time communication network that can connect sensors, actuators and controllers must be established [10,16]. The reasonable scheduling and control of the data uplink of the Internet of things is an indispensable link to ensure the multi service of each user and maintain the stability of the system. How to provide service quality assurance for various businesses and make full use of valuable Internet of things data resources is a problem that must be solved in the dispatching control link.

At present, scholars in related fields have carried out research on network uplink scheduling, and have achieved certain results. Reference [13] proposes a comprehensive research method for uplink scheduling in multi-cell OFDMA networks. We first focus on two scenarios in the homogeneous case and explore how to design efficient and practical uplink schedulers for these scenarios. A centralized multi-cell scheduler is studied to compute the best achievable performance with complete information. The performance of a practical local benchmark scheduler is investigated. This method has certain validity. Reference [14] proposed a low-complexity uplink scheduling algorithm with power control in a wireless logging system based on continuous interference cancellation. By using sequential interference cancellation techniques, the throughput maximization and fairness maximization problems are solved. Aiming at the complexity of these problems, a low-complexity throughput-maximizing approximation algorithm and a low-complexity fairness-maximizing heuristic algorithm are proposed. To evaluate the proposed scheme, extensive simulations were performed. The fairness index of this method can reach 0.975. However, the above methods still have the problems of poor scheduling control effect, low accuracy and low energy efficiency.

To solve the above problems, a scheduling control method for the data uplink of the Internet of things based on TS algorithm is proposed. The specific research ideas are as follows:

Firstly, the principle and model of tabu search algorithm are analyzed, and the characteristics of tabu search algorithm are described. By establishing the objective function and setting the constraints, the scheduling control problem of Internet of things data uplink is described.

Secondly, considering the scheduling control problem of the Internet of things data uplink, independently select the Internet of things equipment base station for link access, and construct the mathematical model of the scheduling control of the Internet of things data uplink.

Then, tabu search algorithm is used to solve the scheduling control problem of Internet of things data uplink, and realize the scheduling control of Internet of things data uplink.

Finally, the uplink scheduling control effect is verified through the experimental indicators of throughput, scheduling control energy efficiency and scheduling control accuracy.

2. Tabu search algorithm

2.1. Principle of Tabu search algorithm

Tabu search (TS) is a meta heuristic algorithm inspired by human memory system [3,9,11]. Its basic idea is as follows: Based on a certain memory pattern, the algorithm will remember several local optimal solutions obtained from the recent search, and treat them as “taboo points” in neighborhood search (this mimics that human beings will not search the searched position when searching for items). In the subsequent search, try to avoid these taboo points (but not absolutely avoid, there is the existence of “amnesty” criterion), so as to ensure different effective search ways and improve the efficiency of local neighborhood search.

2.2. Tabu search algorithm

The Tabu search algorithm involves the following key concepts: initial solution, Tabu table, neighborhood, fitness value function, candidate solution and amnesty criterion [7,8,15].

Initial solution: The quality of the initial solution has a very important impact on the performance and results of the Tabu search algorithm. Under the same other settings, different initial solutions will result in different operating results.

Tabu table and objects: The Tabu table is a table used to remember the selected direction of the neighborhood solution in the recent period. The purpose of this taboo is to avoid roundabout searches as much as possible, so as to find high-quality solutions in the neighborhood space more efficiently.

Neighborhood: The neighborhood structure is generally related to the specific problem to be solved. It is composed of the adjacent subspace of the adjacent solution, the candidate solution set of the adjacent solution, and obtained by the adjacent function.

Adaptation value function: The adaptation value function is a cost function that is ubiquitous in the optimization algorithm, and is used to evaluate the search results.

Candidate solution: a key parameter. When conducting Tabu search, the number of candidate solutions and selection criteria must be determined first, and the number of candidate solutions will have a great impact on the amount of computation and affect the efficiency of the algorithm. Too few candidate solutions lead to premature convergence and premature fall into local minimization.

Amnesty criterion: If all the candidate solutions in a certain neighborhood search result are in the “taboo” state, then the amnesty criterion is triggered at this time.

Termination condition: It is necessary to set a certain criterion as the condition for stopping the algorithm. In fact, there are generally two methods for setting the termination condition: one is to terminate when the solution quality is not improved for n consecutive search results; the other is to artificially specify the number of iterations and stop when the number of iterations reaches that number.

2.3. Tabu search algorithm features

The new Tabu search algorithm is not randomly generated in the vicinity, but is selected from the solutions that satisfy the amnesty condition or the optimal solution in the current iteration process.

Due to the existence of memory and pardon criteria, the algorithm can expand the search range in the neighborhood search, and concentrate the computing power on the effective search. At the same time, the solution set of the algorithm is neutral, that is, the “concentration” is strong. This algorithm is different from ant colony algorithm. Its iterative search program mechanism is continuous, and there is only one state of motion in each iteration. Compared with “concentration”, Tabu search algorithm has weak “exploration” ability and insufficient diversity of solutions generated in the iterative process.

3. Internet of things data uplink scheduling control

Set the constraints, build the scheduling control prediction model of the Internet of things data uplink, output the initial prediction results, optimize the initial prediction results by using the tabu search algorithm, and construct the modified objective function of the neighborhood solution of the exchange operation, that is, turn the project scheduling problem into an optimization problem, and obtain the optimization results of the objective function.

3.1. Internet of things data uplink scheduling control problem description

In order to maximize the energy efficiency of the uplink scheduling control of Internet of things data, this paper defines the ratio of all Internet of things data to the sum of all uplink scheduling control power consumption as the objective function $QW$ . Therefore, the Internet of things data uplink scheduling control problem for optimizing energy efficiency is described as: $\begin{matrix} (1) & QW = \frac{\sum_{n = 1}^{E} \sum_{r = 1}^{R} \sum_{j = 1}^{T / T_{r}} Y_{r}}{\sum_{n = 1}^{E} \sum_{r = 1}^{R} \sum_{j = 1}^{T / T_{r}} \sum_{t = 1}^{T} \sum_{b = 1}^{U} (α_{n, k, j, t, b} \times ω_{n, k, j, t, b})} \end{matrix}$

In formula (1), $α_{n, k, j, t, b}$ is a binary decision variable, indicating that the data block b of the time slot t schedules the upload of the j sampled data that the sensor k of the Internet of things device n controls. $ω_{n, k, j, t, b}$ is the power consumed by the sensor k of the Internet of things device n to the data block b of the time slot t when the j sampled data is uploaded [12]. $Y_{r}$ is the sampling value of the sensor k, and $T_{r}$ is the sampling time interval of the sensor k.

The constraints of the energy efficiency objective function $Q W$ of the Internet of things data uplink scheduling control are:

The transmit power of each Internet of things device n should not exceed the maximum power of the Internet of things device n, expressed as: $\begin{matrix} (2) & ω_{n, k, j} ⩽ ω_{max} \forall n \in E, k \in R, j \in J \end{matrix}$

In formula (2), $ω_{n, k, j}$ is the total power consumed by all data blocks when the sensor k of the Internet of things device n uploads the j sampled data, and $ω_{max}$ is the maximum power of the Internet of things device n.

Ensure that the j data upload of the sensor of the Internet of things device n is scheduled according to the transmit power on all the uploaded data blocks, expressed as: $\begin{matrix} (3) & α_{n, k, j, t, b_{1}} \times ω_{n, k, j, t, b_{1}} = α_{n, k, j, t, b_{2}} \times ω_{n, k, j, t, b_{2}} \forall n \in E, k \in R, j \in J, t \in T, b_{1}, b_{2} \in U \end{matrix}$

The Internet of things data uplink controlled by scheduling should not be less than the data volume of sensor k, which is expressed as: $\begin{matrix} (4) & \sum_{b = 1}^{U} α_{n, k, j, t, b} \times I (β_{n, k, j, t}) ⩾ Y_{r} \end{matrix}$

In formula (4), $β_{n, k, j, t}$ is the modulation and coding strategy index of the j sampling of the sensor k of the Internet of things device n when the data is uploaded at the time slot t, and $I (β_{n, k, j, t})$ is the rate of each data block under the $β_{n, k, j, t}$ index value.

The uplink power control function of Internet of things data uplink scheduling control is expressed as: $\begin{matrix} (5) & ω_{n, k, j, t, b} + 10 \times {log}_{10} (\sum_{b = 1}^{U} α_{n, k, j, t, b}) + (γ - 1) \times O_{n} - δ - P_{0} ⩾ L_{INR} (β_{n, k, j, t}) \end{matrix}$

In formula (5), γ is the path loss compensation coefficient, which is used for Internet of things data uplink scheduling control, $O_{n}$ is the path loss of Internet of things data uplink scheduling control, δ is thermal interference, $P_{0}$ is noise, and $L_{INR} (β_{n, k, j, t})$ is signal and the required thresholds for different modulation and coding methods for interference and noise ratios.

Ensure that the number of data blocks for scheduling control in a single time slot t does not exceed U, which is expressed as: $\begin{matrix} (6) & \sum_{n = 1}^{E} \sum_{r = 1}^{R} \sum_{j = 1}^{T / T_{r}} \sum_{t = 1}^{T} \sum_{b = 1}^{U} α_{n, k, j, t, b} ⩽ U \forall t \in T \end{matrix}$

Multiple data blocks dispatched to a single Internet of things device must be adjacent, expressed as: $\begin{array}{l} (7) & α_{n, k, j, t, b}^{max} - α_{n, k, j, t, b}^{min} = \sum_{b = 1}^{U} α_{n, k, j, t, b} - 1 \\ (8) & α_{n, k, j, t, b}^{max} = max {b | α_{n, k, j, t, b} = 1} \\ (9) & α_{n, k, j, t, b}^{min} = min {b | α_{n, k, j, t, b} = 1} \end{array}$

A data block can only be assigned to one Internet of things device, which is expressed as: $\begin{matrix} (10) & \sum_{n = 1}^{E} \sum_{r = 1}^{R} \sum_{j = 1}^{T / T_{r}} α_{n, k, j, t, b} ⩽ 1 \forall t \in T, b \in U \end{matrix}$

3.2. Internet of things data uplink scheduling control function

This paper considers the scheduling control problem of the data uplink of the Internet of things. The Internet of things equipment can independently select the base station for link access. Assuming that the channel is divided into S sub-channels D, $D = {1, 2, \dots, S}$ , there are coalition bodies F and G in the function, satisfying $F \cap G = \emptyset$ . ε is the ideal signal, ϵ is the interference signal, the power on the θ sub-channel is $P_{ε}^{θ}$ , and the total transmit power is $L_{i}$ . The data uplink scheduling control power vector for Internet of things devices is defined as: $\begin{matrix} (11) & K_{ε} = {K_{ε}^{1}, K_{ε}^{2}, \dots K_{ε}^{θ}} \end{matrix}$

In the θ sub-channels, the uplink carrier-to-interference noise ratio of Internet of things devices under the macro base station Z can be expressed in the following way: $\begin{matrix} (12) & x_{ε ϵ}^{θ} = \frac{c_{ε K}^{θ}}{ϑ + \sum_{ϵ \in U} c_{ϵ F}^{θ} K_{ϵ F}^{θ}} \end{matrix}$

In formula (12), $K_{ϵ F}^{θ}$ is the transmit power of the interference signal ϵ sent from the base station F, and $c_{ε K}^{θ}$ is the channel gain of the K Internet of things device on the θ sub-channel. $c_{ϵ F}^{θ}$ is the channel gain of the interfering signal ϵ from the Internet of things device to the base station F on the θ sub-channel, and ϑ is the power of white Gaussian noise [4].

Using Shannon’s formula to calculate the macro base station Z, the uplink capacity of the Internet of things device in the first sub-slot is expressed as: $\begin{matrix} (13) & v_{ε K} = \frac{M}{S} \sum_{θ \in D} log (1 + x_{ε ϵ}^{θ} \times K_{ε}) \end{matrix}$

In formula (12), M is the Internet of things dynamic bandwidth.

The scheduling control objective of the Internet of things data uplink in this paper is to reasonably schedule and control the power of the Internet of things data uplink on the premise of meeting the scheduling control power constraints of the Internet of things data uplink, so as to maximize the throughput of the Internet of things data uplink [6] as follows: $\begin{matrix} (14) & max T_{t} = maxmin {v_{ε K}} \end{matrix}$

3.3. Realize the scheduling control algorithm of Internet of things data uplink

Based on the above analysis, this paper proposes a Tabu search algorithm to solve the scheduling control problem of Internet of things data uplink. The specific algorithm flow is as follows:

Initial solution and adaptation value function: The initial solution is the optimal solution after the algorithm iteration, and the adaptation function is the maximum energy utilization rate.

Neighborhood search and taboo objects: A swap operation suitable for link scheduling control on the Internet of Things is constructed, and the neighborhood solution is corrected by using the swap operation. Because of its fast operation speed, it can be used as a taboo target in the Tabu search algorithm.

Selection of candidate solutions, Tabu lists and their lengths: The optimal candidate solutions are selected from the generated neighbor solutions, and the group size of $2 GH$ is selected, where the Tabu length is: $\begin{matrix} (15) & T_{B} = round {{[GH (GH - 1) / 2]}^{1 / 2}} \end{matrix}$

Amnesty criterion: A principle based on fitness value is used. If a taboo candidate solution has a better fitness value than “best so far”, the candidate solution will be lifted, the candidate solution is considered to be the current and best, and the “best so far” status will be updated.

Termination criterion: When the iteration of the algorithm reaches a certain maximum, the algorithm will end.

Through the above steps, the scheduling control of Internet of things data uplink is realized. The algorithm implementation process is shown in Fig. 1.

Fig. 1.

Algorithm implementation flow chart.

4. Experimental simulation and analysis

4.1. Setting the experimental environment

In order to verify the effectiveness of the Internet of things data uplink scheduling control method based on the TS algorithm, the TS algorithm was programmed with the MATLAB 2018a software as the platform. And run the TS algorithm on a computer with a performance of Inter Xeon (R) E5-2650 v2 CPU and 2.60 GHz RAM, and the operating system is Windows 7. The number of 25 Internet of things device sensors is selected, and the method of reference [13], the method of reference [14] and the proposed method are used to compare the proposed method to verify the effectiveness of the proposed method.

4.2. Experimental index

Throughput. Throughput refers to the number of requests processed by the system in unit time. The higher the throughput, the better the uplink scheduling control effect of this method. On the contrary, the higher the throughput, the worse the uplink scheduling control effect of this method.

Scheduling controls packet loss rate.

To further verify the accuracy of the proposed method for Internet of things data uplink scheduling control, the scheduling control packet loss rate is used as the evaluation index. The lower the packet loss rate of its scheduling control, the higher the accuracy of the method’s scheduling control of Internet of things data uplink. Its calculation formula is as follows: $\begin{matrix} (16) & D_{B} = \frac{D_{C}}{D_{Z}} \times 100 % \end{matrix}$ In formula (16), $D_{C}$ refers to the number of lost Internet of things data uplink scheduling control packets, and $D_{Z}$ is the total number of Internet of things data uplink scheduling control packets.

Uplink scheduling control energy efficiency. The higher the energy efficiency of Internet of things data uplink scheduling control, the better the uplink scheduling control effect. On the contrary, the lower the energy efficiency of Internet of things data uplink scheduling control, the worse the uplink scheduling control effect.

4.3. Experimental result

4.3.1. Comparison of Internet of things data uplink scheduling control effects

In order to verify the effect of the proposed method on the scheduling control of the Internet of things data uplink, the scheduling control throughput is taken as the evaluation index. The larger the scheduling control throughput, the better the scheduling control effect of the method on the Internet of Things data uplink. The method of reference [13], the method of reference [14] and the proposed method are used to compare, and the comparison results of the Internet of things data uplink scheduling control throughput of different methods are shown in Fig. 2.

Fig. 2.

Comparison results of Internet of things data uplink scheduling control throughput for different methods.

Analysis of Fig. 2 shows that as the number of Internet of things device sensors increases, the throughput of Internet of things data uplink scheduling control for different methods increases. When the number of IOT equipment sensors is 5, the IOT data uplink scheduling control throughput of the method of literature [13] is 2.2 mb/s, and the IOT data uplink scheduling control throughput of the method of literature [14] is 1 MB/s. The scheduling control throughput of the Internet of things data uplink of the proposed method is as high as 4.8 mb/s.When the number of Internet of things device sensors is 25, the Internet of things data uplink scheduling control throughput of the method of reference [13] is 7 Mb/s, and the Internet of things data uplink scheduling control throughput of the method of reference [14] is 3.6 Mb/s. And the proposed method achieves up to 9.4 Mb/s of Internet of things data uplink scheduling control throughput. It can be seen from this that the proposed method has a large throughput of Internet of things data uplink scheduling control, which indicates that the proposed method has a better effect of Internet of things data uplink scheduling control.

4.3.2. Comparison of Internet of things data uplink scheduling control accuracy

The method of reference [13], the method of reference [14] and the proposed method are used to compare, and the comparison results of the packet loss rate of Internet of things data uplink scheduling control of different methods are obtained as shown in Fig. 3.

Fig. 3.

Comparison results of packet loss rate of Internet of things data uplink scheduling control by different methods.

Analysis of Fig. 3 shows that with the increase of the number of Internet of things device sensors, the packet loss rate of Internet of things data uplink scheduling control of different methods increases accordingly. When the number of IOT equipment sensors is 5, the packet loss rate of IOT data uplink scheduling control in the method of literature [13] is 2.0%, and the packet loss rate of IOT data uplink scheduling control in the method of literature [14] is 1%. The packet loss rate of the proposed method is only 0.1%.When the number of Internet of things device sensors is 25, the packet loss rate of the Internet of things data uplink scheduling control method of the method of reference [13] is 2.8%, and the packet loss rate of the Internet of things data uplink scheduling control method of the method of reference [14] is 2.8%. 4.5%. However, the packet loss rate of the proposed method for the uplink scheduling control of Internet of things data is only 1.4%. It can be seen that the packet loss rate of the Internet of things data uplink scheduling control of the proposed method is small, indicating that the proposed method has a high accuracy of the Internet of things data uplink scheduling control.

4.3.3. Energy efficiency comparison of Internet of things data uplink scheduling control

On this basis, verify the energy efficiency of Internet of things data uplink scheduling control of the proposed method, and compare the method of reference [13], the method of reference [14] and the proposed method respectively. The comparison results of energy efficiency of Internet of things data uplink scheduling control of different methods are shown in Table 1.

Table 1
Comparison results of energy efficiency of Internet of things data uplink scheduling control with different methods

Internet of things device sensor quantity/piece The proposed method/bit/J The method of reference [13]/bit/J The method of reference [14]/bit/J

5 5.1 2.3 3.6

10 7.4 3.6 4.1

15 8.2 3.8 5.5

20 9.6 4.4 6.0

25 10.4 5.1 6.2

Internet of things device sensor quantity/piece	The proposed method/bit/J	The method of reference [13]/bit/J	The method of reference [14]/bit/J
5	5.1	2.3	3.6
10	7.4	3.6	4.1
15	8.2	3.8	5.5
20	9.6	4.4	6.0
25	10.4	5.1	6.2

Analysis of Table 1 shows that with the increase of the number of Internet of things device sensors, the energy efficiency of Internet of things data uplink scheduling control of different methods increases accordingly. When the number of IOT equipment sensors is 5, the energy efficiency of IOT data uplink scheduling control in the method of literature [13] is 2.3 bit/j, the energy efficiency of IOT data uplink scheduling control in the method of literature [14] is 3.6 bit/j, and the energy efficiency of IOT data uplink scheduling control in the proposed method is as high as 5.1 bit/j.When the number of Internet of things device sensors is 25, the energy efficiency of the Internet of things data uplink scheduling control method of the method of reference [13] is 5.1 bit/J, and the energy efficiency of the Internet of things data uplink scheduling control method of the method of reference [14] is 6.2 bit/J, while the energy efficiency of the proposed method for Internet of things data uplink scheduling control is as high as 10.4 bit/J. It can be seen that the proposed method has high energy efficiency in the scheduling control of the Internet of things data uplink.

5. Conclusion

The scheduling control method of Internet of things data uplink based on TS algorithm proposed in this paper adopts the Tabu search algorithm to schedule and control the Internet of things data uplink. The method has a better effect of scheduling and controlling the Internet of Things data uplink, and can effectively improve the accuracy and energy efficiency of the Internet of Things data uplink scheduling and control.

The experimental results show that:

When the number of sensors in the Internet of things equipment is 25, the scheduling control throughput of the Internet of things data uplink of the proposed method is as high as 9.4 mb/s. The results show that the scheduling control effect of the proposed method is good.

When the number of sensors in the IOT equipment is 25, the packet loss rate of the proposed method is only 1.4%. It shows that the scheduling control accuracy of the Internet of things data uplink of the proposed method is high.

When the number of sensors in IOT equipment is 25, the energy efficiency of the proposed method is as high as 10.4 bit/j. It shows that the proposed method has high energy efficiency in the scheduling control of the Internet of things data uplink.

But this method is only studied for the simulation environment. Therefore, in the following research, considering the needs of practical applications, the simulation experiment environment is further improved, so as to effectively ensure the scheduling control effect of the Internet of things data uplink.

Footnotes

Acknowledgements

This work was supported by Anhui Provincial Education Department Foundation under grant no.Design of campus lake monitoring system based on Internet of things, and Natural Science research project of Higher education in Anhui Province under grant no. KJ2020A1091.

References

M.J.

Baucas,

Spachos and

Gregori, Internet-of-Things devices and assistive technologies for health care: Applications, challenges, and opportunities, IEEE Signal Processing Magazine 38(4) (2021), 65–77. doi:10.1109/MSP.2021.3075929.

Cheng,

Shi,

Wu,

Guo and

Xiong, An intelligent scheme for big data recovery in Internet of things based on multi-attribute assistance and extremely randomized trees, Information Sciences 557(21) (2021), 66–83. doi:10.1016/j.ins.2020.12.041.

Ghnatios,

R.M.

Hage and

Hage, An efficient tabu-search optimized regression for data-driven modeling, Comptes Rendus Mécanique 347(11) (2019), 806–816. doi:10.1016/j.crme.2019.11.006.

Goswami and

Chakrabarti, Stochastic resetting and first arrival subjected to Gaussian noise and Poisson white noise, Physical Review E 104(3) (2021), 1103.

Hou,

Xu,

Zheng and

W.X.

An, Intelligent transaction migration scheme for RAFT-based private blockchain in Internet of things applications, IEEE Communications Letters 25(8) (2021), 2753–2757. doi:10.1109/LCOMM.2021.3079201.

Khoshkbari,

Pourahmadi and

S.H.

Bayesian, Reinforcement learning for link-level throughput maximization, IEEE Communications Letters 24(8) (2020), 1738–1741. doi:10.1109/LCOMM.2020.2990308.

T.G.

Kim,

Y.R.

Lee,

B.J.

Kang and

E.G.

Im, Binary executable file similarity calculation using function matching, Journal of supercomputing 75(2) (2019), 607–622. doi:10.1007/s11227-016-1941-2.

Lai,

J.K.

Hao,

Glover and

Yue, Intensification-driven tabu search for the minimum differential dispersion problem, Knowledge-Based Systems 167(05) (2019), 68–86. doi:10.1016/j.knosys.2019.01.010.

Liu,

Zhang,

Kurniawan,

Juhana and

Ai, Tabu-search-based pilot assignment for cell-free massive MIMO systems, IEEE Transactions on Vehicular Technology 69(2) (2020), 2286–2290. doi:10.1109/TVT.2019.2956217.

10.

Mukherjee,

Goswami,

Yan and

Yang, Adaptive particle swarm optimisation based energy efficient dynamic correlation behavior of secondary nodes in cognitive radio sensor networks, IET Communications 14(10) (2020), 1658–1665. doi:10.1049/iet-com.2019.0750.

11.

N.T.

Nguyen and

Lee, Deep learning-aided Tabu search detection for large MIMO systems, IEEE Transactions on Wireless Communications 19(6) (2020), 4262–4275. doi:10.1109/TWC.2020.2981919.

12.

Ono and

Y.M.

Power, Consumption analysis of optical repeater subsystem in multicore fiber link, Journal of Lightwave Technology 39(14) (2021), 4629–4637. doi:10.1109/JLT.2021.3077651.

13.

Ozcan and

R.C.

Uplink, Scheduling in multi-cell OFDMA networks: A comprehensive study, IEEE Transactions on Mobile Computing 20(10) (2020), 3081–3098. doi:10.1109/TMC.2020.2994354.

14.

Xu,

Ding and

Xu, Low-complexity uplink scheduling algorithms with power control in successive interference cancellation based wireless mud-logging systems, Wireless Networks 25(1) (2019), 321–334. doi:10.1007/s11276-017-1561-7.

15.

Yin, Simulation of accurate extraction for multiple candidate results based on semi-supervised learning, Computer Simulation 37(8) (2020), 435–438.

16.

Zhang,

Han,

Wang,

Guizani,

Fan and

Shu, A node location algorithm based on node movement prediction in underwater acoustic sensor networks, IEEE Transactions on Vehicular Technology 69(3) (2020), 3166–3178. doi:10.1109/TVT.2019.2963406.