1. bookAHEAD OF PRINT
Detalles de la revista
License
Formato
Revista
eISSN
2444-8656
Primera edición
01 Jan 2016
Calendario de la edición
2 veces al año
Idiomas
Inglés
Acceso abierto

Label big data compression in Internet of things based on piecewise linear regression

Publicado en línea: 15 Jul 2022
Volumen & Edición: AHEAD OF PRINT
Páginas: -
Recibido: 14 Mar 2022
Aceptado: 27 May 2022
Detalles de la revista
License
Formato
Revista
eISSN
2444-8656
Primera edición
01 Jan 2016
Calendario de la edición
2 veces al año
Idiomas
Inglés
Introduction

With the increasing maturity of wireless communication, integrated circuit, sensor and micro electro mechanical system (MEMS), it is possible to produce micro wireless sensors with low cost, low power consumption and multi-function. These sensor nodes integrate functional components such as information sensing, data processing and wireless communication [1]. WSN is a wireless self-adjusting network composed of multiple miniature smart sensor nodes sent to the monitoring area. Its purpose is to monitor, perceive and collect the data of various environments or monitoring objects in the geographical area covered by the network in real time, process these data, obtain detailed and accurate information and transmit them to users who need these information. With the help of various sensor units built in the node, the signals of heat, infrared, magnetism, image, sonar, radar and seismic wave in the surrounding environment are measured, so as to detect many physical phenomena of interest, including temperature, humidity, noise, light intensity, pressure, soil composition, size, speed and direction of moving objects. WSN is usually deployed in extreme environments that cannot be reached or stationed for a long time, such as the field or disaster site. Widely used in national defense and military industry, anti-terrorism and disaster prevention, smart home, environmental protection, earthquake and extreme weather, risk control, vehicle control, medical maintenance, manufacturing and other fields. This is currently very active research. Cai M et al. regard wireless sensor networks as a new technology for information understanding, detection and processing. If the Internet recognizes that the world communicates for digital purposes and changes the way people communicate; then, WSN recognizes the connection between the digital world and the physical goals of the world, changes the interaction of people and events, and marks the coming of counting time[2]. Zhang J and others found that WSN originated from the combat needs of the U.S. military. Since the mid-1990s, WSN has attracted great attention from the military, academia and industry. The United States has invested heavily in supporting the research of WSN through the Advanced Research Projects Agency (DARPA) of the Ministry of defense and the National Natural Science Foundation (NSF) [3]. Typical research projects include DARPA's SensIT(Sensor Information Technology) research plan, which includes 29 research projects. At present, some sensor nodes and operating systems that can be put into practical use have been developed. Representative sensor nodes are mica series and Imote2 series produced by crossbow company. Ti, Freescale and other companies have developed ZigBee series chips such as CC2430. Famous operating systems include TinyOs developed by UCB and SOS developed by UCLA, but SOS has not been upgraded since 2008. In recent years, the National Natural Science Foundation of China, 863 program and 973 program have provided large-scale and high-intensity research funding in the field of WSN, which has accelerated the research process of WSN in China. In China, the earlier institutions to carry out WSN related research include Institute of computing, Institute of software, Chinese Academy of Sciences, Harbin Institute of technology, Tsinghua University, Shanghai Jiaotong University, Peking University, Nanjing University, National Defense University, Zhejiang University, Fudan University, Beijing University of Posts and telecommunications, Central South University, Hunan University, etc. At the beginning of 2006, 973 Plan established the basic theory and key technology research project of wireless sensor networks. More than ten key universities, including Hong Kong University of science and technology and Shanghai Jiaotong University, jointly carried out research on all levels of sensor networks. Ningbo Zhongke integrated circuit design center (NBICC) was established by the Institute of computing of the Chinese Academy of Sciences in cooperation with Ningbo municipal government. It specializes in providing solutions and system integration based on radio frequency identification system (RFID), wireless network and electronic tag equipment. Ningbo high tech Zone Shenlian Technology Co., Ltd. is a high-tech enterprise incubated by NBCC wireless communication division, specializing in short-range wireless communication such as WSN and RFID, embedded communication solutions, etc. Shenlian technology has been engaged in the research and development of wireless communication, embedded, especially WSN. It is a leader in the industrialization of WSN in China. Its gain series nodes and development kits have a high market share in China. The Chinese government has always attached importance to the development of sensor networks. The outline of the national medium and long term science and technology development plan released in early 2006 identified three cutting-edge directions for information technology, two of which are directly related to WSN research. In November 2008, IBM proposed “smart earth”, whose core view is to integrate the Internet of things and the Internet to realize the integration of human society and the physical world. Chen Jie and others believe that the so-called Internet of Things refers to a large-scale network integrated with various data measurement devices, such as RFID, infrared sensors, earthmoving systems, laser scanners, and other devices connected to the Internet or telephone communication[4]. Put sensors on everything that has nothing to do with the network but is closely related to our life and work, and then connect with the existing network, so that people can identify, control and manage these things more directly through computers or mobile phones. In August 2009, Premier Wen Jiabao of the State Council said during his visit to Wuxi Gaoxin micro nano sensor network engineering technology research and development center of the Chinese Academy of Sciences that at least three things can be done as soon as possible: first, combine the sensor system with TD technology in 3G; Second, accelerate the development of sensor networks in major national science and technology projects; Third, establish China's sensing information center, or “sensing China” center, as soon as possible. In September 2009, the National Information Technology Standardization Technical Committee established the working group on sensor network standards, which means that China's “Internet of things” standard has been preliminarily established. One of the main tasks of wireless sensor networks is to collect data in the monitoring area. Sensor nodes have limited energy, bandwidth, computing and storage capacity. How to achieve low energy consumption and high security data collection in resource constrained and node intensive WSN is an important problem that researchers need to solve. Because sensor nodes usually have a certain data processing capacity, they can process the original sensor data in the network through the cooperation between sensor nodes, and then transmit the processing results to sink nodes, so as to save data transmission energy consumption, reduce bandwidth requirements and improve the efficiency of data collection. At the same time, in the process of data transmission, security issues such as data confidentiality, source authentication, integrity and freshness should be considered [5]. Aiming at energy, delay, security and storage efficiency, this paper focuses on the data collection in wireless sensor networks, focusing on the data compression algorithm and data authentication mechanism in the process of data collection from sensor collection site to sink.

Method
P - M model

Wireless sensor networks consist of many affordable tiny sensors. It creates personal organization through wireless multi-hop communication. It understands, compiles and records the information found within the coverage area and sends it to the inspector. Network architecture of wireless sensor networks [6]. Sensing object, sensor and observer constitute the three elements of sensor network. In addition to information collection and data processing, each node in the network also undertakes the storage, management and integration of forwarded data. At the same time, it has the dual functions of terminal and routing. At present, the software and hardware technology of sensor nodes is the research focus of sensor networks. Wireless sensor networks are composed of sensor nodes distributed in specific areas, which are used to monitor specific objects such as temperature, humidity, vibration and so on. A single node in wireless sensor networks is generally composed of sensor module, microprocessor module, wireless communication module and energy supply module.

Although the sensor nodes in the network accomplish special data and network communication, there are still some limitations and limitations.

Limited node energy supply

In order to meet the portable and mobile application requirements of sensor nodes, nodes are generally powered by batteries with very limited energy. Sensor nodes are usually delivered randomly, the deployment area environment is complex, and even some area personnel cannot reach, so it is difficult to supplement energy by battery replacement [7]. Therefore, energy is an extremely important and limited resource in wireless sensor networks, and it is also the main constraint of nodes.

The energy consumption of wireless communication is relatively large

The energy consumption of sensor nodes mainly comes from microprocessor module, sensor module and wireless communication module. With the progress of integration technology and microelectronics technology, the energy consumption of microprocessor and sensor is getting lower and lower, so that the wireless data communication part consumes most of the energy of the system. The energy consumption of sensor nodes is as follows. Facing the huge communication cost, how to reduce the amount of communication data has become the key to reduce energy consumption.

Node computing and storage resources are limited

Nodes in wireless sensor networks are usually low-cost and low-power micro embedded devices, and the processor capacity and memory capacity will be limited. On the other hand, due to the limitation of energy, nodes cannot use high-performance and high-power processors. How to use limited computing and storage resources to complete specific tasks has become a challenge in the process of node software design.

Limited wireless communication capability

The distance of wireless communication is the most important factor to determine the communication energy consumption. In addition, the mirror reflection of obstacles, antenna quality and other factors will also affect the communication distance, and the energy consumption of nodes will increase significantly with the increase of wireless communication distance. In order to reduce energy consumption, the communication distance should be minimized on the premise of meeting the connectivity requirements. On the other hand, the communication bandwidth of sensor nodes is also very limited, usually only a few hundred Kbps. From the above limitations faced by sensor nodes, it can be seen that energy limitation is the main limitation, and other limitations also come from the limitation or correlation of energy. How to use limited energy and limited bandwidth to complete the transmission of a large amount of data under the condition of limited computing and storage resources, reduce the energy consumption of nodes and prolong the life cycle of equipment will be the biggest challenge faced by wireless sensor networks [8].

Given that the communication power consumption of sensor nodes is higher than that of processors, reducing the cost of data communication will be an important way to reduce system power consumption. Among them, data compression is one of the very effective methods, which has become the research hotspot of energy-saving methods in wireless sensor networks. Because the node monitoring data usually changes slowly in a short time, the data may be relevant in time; Nodes are densely deployed in the monitoring area, and multi-point cooperation may be adopted, so that the data perceived by adjacent nodes may also be spatially relevant. In addition to the temporal and spatial correlation, the monitoring data may also show certain temporal and spatial regularity, such as certain trend, distribution characteristics or repeatability. These temporal and spatial correlation and regularity of the data provide the possibility for data compression. Compared with the emerging wireless sensor networks, data compression technology, as an important research field in Information Science, has a long history of development [9]. The combination of data compression technology and wireless sensor network to reduce the amount of communication data not only reduces the energy consumption of nodes, but also effectively reduces the network congestion. For the whole network, it improves the data transmission efficiency and bandwidth utilization, and reduces the energy consumption of the whole network, as shown in Table 1.

Classification of resource constraints of hardware platform

Platform Power supply Storage space Processing capacity
personal computer infinite Unrestricted Unrestricted
High end embedded Limited Relatively limited Relatively limited
Low end embedded Extremely limited Extremely limited Extremely limited

The resources of the hardware platform determine the scale of the running algorithm. According to the different constraints of the hardware platform resources (energy supply, storage space and processing capacity), the application platforms of the current algorithm can be divided into several categories as shown in Table 1. For general wireless sensor network nodes, not only the power supply is extremely limited, but also the storage space and processing capacity are limited. However, traditional data compression algorithms usually run on personal computers with almost unlimited resources, and take the compression rate as the measurement standard of the algorithm. When these compression algorithms are introduced into wireless sensor networks, while pursuing high compression rate to reduce communication energy consumption, we must also consider the resource constraints of WSN, and take the energy-saving benefit of the algorithm as the measurement standard. Therefore, this also puts forward new requirements for the compression algorithm.

Energy efficiency

Energy saving is the most important purpose of applying data compression technology to wireless sensor networks, that is, reducing the energy consumption of nodes by reducing the amount of communication data. At the same time, the operation of compression algorithm on hardware increases the energy consumption of nodes. Therefore, in order to ensure that the running compression algorithm can bring a certain energy-saving effect to the node, it must be required that the energy consumed by running the compression algorithm should be less than the transmission energy of sending the reduced amount of data through compression, otherwise the significance of data compression will be lost.

Low algorithm complexity

As can be seen from table 1, in addition to the limited energy, the calculation and processing capacity of sensor nodes is also limited, which is mainly reflected in the storage space and processing capacity. Therefore, it can not bear or realize a large number of complex operations. The limited computing capacity of node microprocessor requires that the time complexity of the algorithm should be as low as possible; The storage medium with limited nodes requires that the amount of dynamic data of the algorithm cannot be too much. For example, K-L transform, nonlinear prediction, subband coding and so on will not be directly introduced into the sensor node.

High execution efficiency

The compression algorithm running on the sensor node finally needs the source program to realize. The low time complexity of the compression algorithm itself is the premise of efficient algorithm execution. However, due to the subjectivity of the algorithm source program in the programming process, there will be some differences in the execution efficiency of the algorithm source program realized by different implementation methods [10]. Therefore, in addition to considering the complexity of the algorithm, the data compression algorithm in wireless sensor networks also needs to consider the efficiency of the algorithm itself. Appropriate optimization methods should be introduced in the process of source program design to improve the execution efficiency of the algorithm.

Dynamic

The design of data compression algorithm usually aims at specific data objects, that is, the algorithm can obtain good compression effect under a specific data condition. For example, LZW is suitable for text data compression, run length coding is suitable for compressing a large number of continuous character data, etc. When the data characteristics change, the compression effect will change significantly. The environment detected by sensor nodes is usually unknown and the data characteristics are unstable. In order to obtain good compression effect, the algorithm should have fixed dynamic characteristics and be able to dynamically adapt to the changes of environment and data characteristics.

For the traditional applications of data compression, compression algorithms mainly focus on using data compression technology to reduce the amount of data, so as to effectively reduce the storage space of data, reduce the time and transmission bandwidth of data communication. When these algorithms are improved and introduced into wireless sensor networks, some original evaluation indexes of algorithms are also introduced. At present, several common indicators for evaluating the performance of compression algorithms are as follows:

The data collected by sensor nodes can be described as a matrix structure, as shown in formula (1) where the row vector represents a single parameter to sample the dataset over the time series. The n × m matrix represents the current latest sample data, n represents that the node has n monitoring parameters, and m represents the number of sampling periods of the node (assuming that each parameter has the same sampling period): X=[Y1Y2Yn]=[y11y12y1my21y22y2myn1yn2ynm] X = \left[ {\matrix{{{Y_1}} \cr {{Y_2}} \cr \ldots \cr {{Y_n}} \cr}} \right] = \left[ {\matrix{{{y_{11}}} & {{y_{12}}} & \ldots & {{y_{1m}}} \cr {{y_{21}}} & {{y_{22}}} & \ldots & {{y_{2m}}} \cr {} & {} & \ldots & {} \cr {{y_{n1}}} & {{y_{n2}}} & \ldots & {{y_{nm}}} \cr}} \right]

For strongly correlated parameters (such as air temperature and soil temperature), a standard linear regression estimation model is used to establish a mathematical relationship equation between the parameters, as shown in (2), and the least squares method is used to determine the fitting equation when the fitting error RMSE is the smallest. Regression coefficients (a, b). The sensor network node sends Y; as the reference data set BD (Base Data) to the base station, and the transmission of the Y; data set can be realized by only sending the regression coefficients (a, b) of the fitting equation, see equation (2). Yi=aYj+b+Э(1<i,j<=n) {Y_i} = a \bullet {Y_j} + b + \ni \left({1 < i,j < = n} \right)

However, performing linear regression on the entire time series will produce relatively large errors, mainly due to: (1) The correlation degree of parameters in different time periods is different. For example, the rate of change of ambient temperature is small in the noon time of the day, while the rate of change is large in the evening. ② The interaction of different parameters has a lag in the time dimension. When a parameter changes, the corresponding changes in other related parameters are delayed. For example, the interaction between ambient temperature and soil temperature. A piecewise linear regression mapping method is proposed, which divides the sample data on the time series into segments, performs linear regression calculation in units of segments, and performs feature description in the form of a self-defined structured reduced set S, as defined in Definition 1. After the n × m sample data collection is completed, the sensor network node first selects the reference data set (set as yj), and divides the row i into [m / l] segments, [m / l] indicates that the calculated value of m / l is rounded up. Each segment searches for the optimal data interval with a length of yj; on the corresponding linear regression operation, and sends the result to the base station in a structured form. The entire reference data set is directly sent to the base station as the basis for data recovery. After the segmented regression operation is completed for each row, the row can be replaced by a structured reduced set Sk (Sk << m, k = [m / l]), and the reduced set is used as the feature description of row i to perform data recovery at the base station. Assuming that yj is the benchmark data set, a certain segment of data Yi in Yi [start, start + l] can be expressed as formula (3): aYj[migration,migration+l]+b a \cdot {Y_j}\left[ {migration,\,\,migration + l} \right] + b

Assuming that the data representation of the sensor acquisition parameters and the data representation of the S-member parameters of the structured and reduced set occupy the same storage space, the data compression rate can be simplified as shown in Equation (4): CR=1i=1,ijnmli4+mnm CR = 1 - {{\sum\limits_{i = 1,\,i \ne j}^n {{m \over {{l_i}}} \cdot 4 + m}} \over {n \cdot m}}

In the formula: li is the i segment length of the eighth parameter, n is the number of monitoring parameter types of the sensor network node, and m is the number of sampling periods of the node.

Algorithm complexity

The complexity of an algorithm is an important index to reflect the quality of an algorithm. In data compression algorithms, algorithm complexity actually refers to the hardware resources required in the compression process, usually including the time complexity and space complexity of the algorithm, that is, the amount of computation and storage required by the algorithm. Time complexity refers to the relationship between time frequency T(n) and the scale n of the algorithm problem, which is recorded as T[n] = O(f(n)), which is used to evaluate the time performance of an algorithm; Similar to the time complexity of the algorithm, the spatial complexity of the algorithm refers to a measure of the storage space required by the algorithm during operation, which is recorded as S(n) = O(f(n)), including the storage space occupied by the algorithm source program, the storage space occupied by the initial data and the storage space required during execution [11].

Experiment and analysis

In the process of evaluating the energy efficiency of data compression, in order to make the analysis not stick to the specific environment and specific algorithm, the hardware factors and algorithm factors are discussed separately. The characteristics of typical sensor nodes' hardware environment and typical data compression algorithm are shown in Table 2 respectively.

Hardware structure of typical sensors

Node platform Organization / manufacturer Microprocessor RF chip
Mica2 UCB AT mega128L CC1000
Micaz UCB AT mega128L CC2420
Toles UCB TI MSP430F149 CC2420
T-mote sky Moteiv TI MSP430F2611 CC2420
ZebraNET Princeton TI MSP430F149 9XStream
EyesIFX EURO TI MSP430F149 TDA5250
TinyNode TinyNode TI MSP430F2618 XE1205
Fleck3 SCIRO AT mega128L nRF905
Imote2 Intel Intel PXA271 XScale CC2420
u AMPS-I MIT SA1100 LMX3162

It can be seen from Table 2 that the microprocessors used on typical sensor nodes mainly include ATmegl28L, MSP430 series, PXA271, SA11000, etc. their average current consumption under different working environments is shown in Table 3:

Operating characteristics and average current of common microprocessors on sensor nodes

Microprocessor Working voltage Operating frequency Average current consumption
ATmega128 2.7V 1MHz 1.9mA
3.3V 1MHz 2.1mA
2.7V 8MHz 7.5mA
3.3V 8MHz 9. 5mA
MSP430F2611MSP430F261 8 2.2V 4KHz 2.1uA
3.0V 4KHz 3.0uA
2.2V 1MHz 365uA
3.0V 1MHz 512uA
MSP430F149 2.2V 4KHz 2.5uA
3.0V 4KHz 9.0uA
2.2V 1KHz 280uA
3.0V 1KHz 420uA
ML67Q5002 2.5V 60MHz 75mA

The efficiency and low complexity of the software algorithm itself is the key to reduce the system energy consumption, but under the premise of the same algorithm, the efficiency of the algorithm implementation process source program is also very important [12]. High level language not only simplifies the development process of the algorithm, but also reduces the efficiency of program execution. At the same time, it makes the coding efficiency of different programmers different. Therefore, there is a lot of room for improvement in the optimization of software energy consumption at the source program level. In addition, for current microprocessors, there is a common feature, that is, the capacity of program memory is often much larger than that of data memory, and there are few internal registers. Although the microprocessor is also improving in hardware and the capacity of program memory is increasing, the capacity of data memory (RAM) and the number of internal registers change very little. This resource is relatively short, which also makes it possible to optimize the source program. As the most widely used high-level language, C language has high operation efficiency and good portability, and is often used in the development of underlying programs of embedded systems [13]. Wireless sensor network node is a low-end embedded hardware platform, and C language is the most commonly used development language. Therefore, the research on the algorithm source program level energy consumption optimization method of C language is of great significance to reduce the system operation energy consumption. The program is composed of data structure and algorithm. Different data organization forms and data types will correspond to different access methods and data formats. Both of them determine the execution efficiency and corresponding energy consumption level of the program to a great extent. Adjusting the storage and distribution mode of data in the source program of data compression algorithm in wireless sensor networks is of great significance to reduce the energy consumption of the algorithm.

In C language, all statements are executed in the order of top-down. Even statements or expressions with the same status in structure have the order of execution. For example, the logical operation of each branch of the selection structure and multiple relational expressions, and the sequence of the selected branches and relational expressions do not affect the functions realized by the source program. When the probability distribution of the execution of these branches with the same status is uneven, or the probability of the establishment of the expression is different, the sequence of executing statements or expressions will affect the execution efficiency of the source program. Therefore, when these probabilities are known, adjusting the order of statements or expressions is conducive to improve the execution efficiency of the source program and reduce the system energy consumption. The adjustable range of the hardware coefficient of the sensor node is [111, 227], so the algorithm adjustment points within this adjustment range are only kwch-S- lz W. b-RLA, that is, only [S-LZW, b~RLE] algorithm combination meets the requirements. Through the above test, the algorithm combination for algorithm adjustment and the k value of adjustment point can be obtained. Applying the combination of this algorithm and the values of adjustment points to the lossless compression of node data can realize the data compression with power consumption perception, so that the energy efficiency of nodes can be dynamically maintained at a high level [14].

Conclusion

As an important energy-saving means, data compression is a research hotspot in wireless sensor networks. The research on data compression in wireless sensor networks mainly involves two problems: one is how to objectively and comprehensively evaluate the compression algorithm and establish the corresponding evaluation index; The second is how to improve the energy-saving effect of compression algorithm and prolong the life cycle of nodes. Since energy saving is the primary purpose of WSN data compression, in addition to the traditional evaluation indexes, the relevant evaluation standards should be energy related, focusing on the energy-saving effect of the algorithm; In addition, because the development process of data compression is much earlier than that of wireless sensor networks, and the algorithm has been relatively mature, there is little space to improve the performance of compression by improving the algorithm itself, so we need to rely on other auxiliary methods to further improve the performance of the algorithm. To solve the above problems, focusing on the energy efficiency of compression algorithm, this paper involves two main parts: energy efficiency evaluation index and energy efficiency improvement method. The innovations and contributions of this paper are as follows: Based on the existing energy efficiency evaluation indexes and the characteristics of wireless sensor networks, an improved scheme of energy efficiency evaluation is proposed. The scheme separates the hardware factors and algorithm factors in the original energy efficiency index, makes the evaluation process more flexible and more suitable for wireless sensor networks, simplifies the original evaluation index, and the new scheme is more conducive to the analysis of the influencing factors of energy efficiency.

Operating characteristics and average current of common microprocessors on sensor nodes

Microprocessor Working voltage Operating frequency Average current consumption
ATmega128 2.7V 1MHz 1.9mA
3.3V 1MHz 2.1mA
2.7V 8MHz 7.5mA
3.3V 8MHz 9. 5mA
MSP430F2611MSP430F261 8 2.2V 4KHz 2.1uA
3.0V 4KHz 3.0uA
2.2V 1MHz 365uA
3.0V 1MHz 512uA
MSP430F149 2.2V 4KHz 2.5uA
3.0V 4KHz 9.0uA
2.2V 1KHz 280uA
3.0V 1KHz 420uA
ML67Q5002 2.5V 60MHz 75mA

Classification of resource constraints of hardware platform

Platform Power supply Storage space Processing capacity
personal computer infinite Unrestricted Unrestricted
High end embedded Limited Relatively limited Relatively limited
Low end embedded Extremely limited Extremely limited Extremely limited

Hardware structure of typical sensors

Node platform Organization / manufacturer Microprocessor RF chip
Mica2 UCB AT mega128L CC1000
Micaz UCB AT mega128L CC2420
Toles UCB TI MSP430F149 CC2420
T-mote sky Moteiv TI MSP430F2611 CC2420
ZebraNET Princeton TI MSP430F149 9XStream
EyesIFX EURO TI MSP430F149 TDA5250
TinyNode TinyNode TI MSP430F2618 XE1205
Fleck3 SCIRO AT mega128L nRF905
Imote2 Intel Intel PXA271 XScale CC2420
u AMPS-I MIT SA1100 LMX3162

Zhao Z, Wang H, Wang J, et al. Research on fault repair method of all-optical network based on SDN[J]. China Communications, 2020, 17(6):180–195. ZhaoZ WangH WangJ Research on fault repair method of all-optical network based on SDN[J] China Communications 2020 17 6 180 195 10.23919/JCC.2020.06.015 Search in Google Scholar

Cai M, Wang S, Wu C. Research on real-time data transmission and multi-scale video image decomposition of embedded optical sensor array based on machine learning[J]. Multimedia Tools and Applications, 2020(6):1–21. CaiM WangS WuC Research on real-time data transmission and multi-scale video image decomposition of embedded optical sensor array based on machine learning[J] Multimedia Tools and Applications 2020 6 1 21 10.1007/s11042-020-09847-w Search in Google Scholar

Zhang J. Interaction design research based on large data rule mining and blockchain communication technology[J]. Soft Computing, 2020, 24(21):16593–16604. ZhangJ Interaction design research based on large data rule mining and blockchain communication technology[J] Soft Computing 2020 24 21 16593 16604 10.1007/s00500-020-04962-0 Search in Google Scholar

Chen J, Zhao F, Xing H. Research on Security of Mobile Communication Information Transmission Based on Heterogeneous Network[J]. International Journal of Network Security, 2020, 22(1):145–149. ChenJ ZhaoF XingH Research on Security of Mobile Communication Information Transmission Based on Heterogeneous Network[J] International Journal of Network Security 2020 22 1 145 149 Search in Google Scholar

Ji B, Han Y, Li P, et al. Research on Secure Transmission Performance of Electric Vehicles Under Nakagami-m Channel[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, PP(99):1–11. JiB HanY LiP Research on Secure Transmission Performance of Electric Vehicles Under Nakagami-m Channel[J] IEEE Transactions on Intelligent Transportation Systems 2020 PP 99 1 11 10.1109/TITS.2020.3030540 Search in Google Scholar

Li B, Li Y E, Yang J. Q-interface imaging using accumulative attenuation estimation[J]. Geophysics, 2020, 85(6):1–85. LiB LiY E YangJ Q-interface imaging using accumulative attenuation estimation[J] Geophysics 2020 85 6 1 85 10.1190/geo2019-0759.1 Search in Google Scholar

Guo L. Research on Anomaly Detection in Massive Multimedia Data Transmission Network Based on Improved PSO Algorithm[J]. IEEE Access, 2020, PP(99):1–1. GuoL Research on Anomaly Detection in Massive Multimedia Data Transmission Network Based on Improved PSO Algorithm[J] IEEE Access 2020 PP 99 1 1 10.1109/ACCESS.2020.2994578 Search in Google Scholar

Liang H, Zuo C, Wei W. Detection and Evaluation Method of Transmission Line Defects Based on Deep Learning[J]. IEEE Access, 2020, 8:38448–38458. LiangH ZuoC WeiW Detection and Evaluation Method of Transmission Line Defects Based on Deep Learning[J] IEEE Access 2020 8 38448 38458 10.1109/ACCESS.2020.2974798 Search in Google Scholar

Hou R, Liu H, Hu Y, et al. Research on secure transmission and storage of energy IoT information based on Blockchain[J]. Peer-to-Peer Networking and Applications, 2020, 13(4):1225–1235. HouR LiuH HuY Research on secure transmission and storage of energy IoT information based on Blockchain[J] Peer-to-Peer Networking and Applications 2020 13 4 1225 1235 10.1007/s12083-019-00856-7 Search in Google Scholar

Xu Z, Shi D, Tu Z. Research on Diagnostic Information of Smart Medical Care Based on Big Data[J]. Journal of Healthcare Engineering, 2021, 2021(10):1–10. XuZ ShiD TuZ Research on Diagnostic Information of Smart Medical Care Based on Big Data[J] Journal of Healthcare Engineering 2021 2021 10 1 10 10.1155/2021/9977358819220534188793 Search in Google Scholar

Cai W, Yao H. A Secure Transmission Method of Network Communication Data Based on Symmetric Key Encryption Algorithm[J]. Wireless Personal Communications, 2021(4):1–12. CaiW YaoH A Secure Transmission Method of Network Communication Data Based on Symmetric Key Encryption Algorithm[J] Wireless Personal Communications 2021 4 1 12 10.1007/s11277-021-08266-w Search in Google Scholar

Tang K, Ge M, Dong S, et al. Transmission Contingency Analysis Based on Data-Driven Equivalencing of Radial Distribution Networks Considering Uncertainties[J]. IEEE Access, 2020, 8(99):1–1. TangK GeM DongS Transmission Contingency Analysis Based on Data-Driven Equivalencing of Radial Distribution Networks Considering Uncertainties[J] IEEE Access 2020 8 99 1 1 10.1109/ACCESS.2020.3045697 Search in Google Scholar

Kanna M, Kumar R P, Nandappa S, et al. On Solutions of Fractional order Telegraph Partial Differential Equation by Crank-Nicholson Finite Difference Method[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(2):85–98. KannaM KumarR P NandappaS On Solutions of Fractional order Telegraph Partial Differential Equation by Crank-Nicholson Finite Difference Method[J] Applied Mathematics and Nonlinear Sciences 2020 5 2 85 98 10.2478/amns.2020.2.00017 Search in Google Scholar

Modanli M, Ali Akgül. On Solutions of Fractional order Telegraph Partial Differential Equation by Crank-Nicholson Finite Difference Method[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(1):163–170. ModanliM AkgülAli On Solutions of Fractional order Telegraph Partial Differential Equation by Crank-Nicholson Finite Difference Method[J] Applied Mathematics and Nonlinear Sciences 2020 5 1 163 170 10.2478/amns.2020.1.00015 Search in Google Scholar

Artículos recomendados de Trend MD

Planifique su conferencia remota con Sciendo