Information Fusion Method of Power Internet of Things Based on Low-Voltage Power Line and Micro-Power Wireless Communication

In order to improve the coverage and reliability of the information perception layer of the power Internet of things, a method of constructing a cross layer fusion network of low-voltage power line and micro-power wireless communication (CPW) is proposed. Firstly, a unified medium access control (MAC) layer model of CPW is established to provide basic support for realizing the integration of CPW network layers; then an improved ant colony algorithm combining Brownian motion and local convergence times control is proposed to complete the networking process of CPW; Distribute the sub-service flow of CPW, and propose the bit error rate requirement factor in the service distribution, and accomplish the cross-layer integration of low-voltage power line and micro-power wireless communication network. The simulation results show that the communication link service quality of the cross-layer fusion network is better than that of the power line and wireless dual-mode or cascaded communication networks. The communication efficiency of CPW is improved to 398.20kbits/s. The coverage of the power line and wireless communication fusion network can be expanded by setting corresponding bit error rate demand factors according to different services. It takes into account the balance of communication link quality and network load.


I. INTRODUCTION
Building a widely interconnected power Internet of Things and deeply integrating it with a strong smart grid is an important measure to accomplish the energy Internet integration in the future [1], [2]. The basic guarantee factor for the construction of the power Internet of Things is the comprehensive perception and interconnection of information [3], [4], [5], which should have the characteristics of broad coverage [6], efficient transmission efficiency, and reliable communication quality assurance [7]. The microgrid can be monitored based on the wireless network, of which the estimated performance affected by the wireless network parameters [8].
The associate editor coordinating the review of this manuscript and approving it for publication was Cong Pu .
The MAC layer communication protocol of CPW is designed based on the PRIME protocol of power line communication [28] and the IEEE802.15.4 standard of wireless communication [29]. Power IoT covers a large number of power equipment and sensor nodes, which can be connected to the network through low-voltage power lines and micro-power wireless communication [30], [31]. The effective transmission of CPW signals is ensured by establishing relay nodes. CPW is a star structure, using a hub or switch as the central node of the network. The nodes in the CPW are distributed in a star shape, and the core switch is called the central node. The MAC layer of CPW adopts carrier sense multiple access with collision avoid (CSMA/CA ) and time division multiple access (TDMA) with collision avoidance, which can avoid transmission collision and improve communication reliability. It divides time into several beacon periods by CPW, and its specific time slot division is shown in Figure 1. It can be seen that the beacon frame is sent from the beginning of the beacon period, and the synchronization of network time and information between nodes can be achieved by sending the networking beacon. In the competitive access time slot of the beacon period, the central node and other sub-nodes access the channel through the CSMA/CA mode to complete the CPW networking and data transmission. In the non-contention access time slot of the beacon period, the node accesses the channel through TDMA, and transmits a service with a selected priority. The nodes with certain special properties access the channel through competition in the bound CSMA/CA time slot. The beacon period for CPW medium and low voltage power line and micropower wireless communication has the same time slot division.
The optimal path networking method for CPW accomplishs the integration of low-voltage power line and micro-power wireless communication at the network layer, and the communication at the network layer must be supported by the MAC layer. The frame header and the frame end respectively identify the start and end of the data frame. The frame control code is used to identify the type of the data frame. The network type identifies the transmission network of the data frame. The source node and the destination node are the numbers of the data sending and receiving nodes respectively. The relay routing table records the number of the relay node passed between the source node and the destination node. The bit error rate requirement factor represents the Quality of Service (QoS) requirement of the service. The data length and data fields are used to identify the information of the data to be sent. The cyclic redundancy check (CRC) check is responsible for checking all bytes of data.
In order to improve the reliability of the communication between the central node and other sub-nodes, an optimal communication path between each sub-node and the central node is constructed. The parameters such as delay, communication rate and bit error rate are usually used to evaluate network performance in the communication network of CPW. There may be multiple relay nodes in the communication path from the source node to the destination node. The number of relay nodes is called routing hops. These factors determine whether the selected communication path is the optimal communication path. The communication network topology is represented by a graph using the knowledge of graph theory. The weights of its edge sets are the delay, bit error rate, communication rate and routing hops of the lowvoltage power line communication link and the micro-power wireless communication link. Suppose a complete communication path is R(v s , v d ), there are n nodes in the path, the starting node is v s . The destination node is v d , then the service quality parameter of the communication path where, when N is taken as P, W , and C, it corresponds to the low-voltage power line communication network, the micro-power wireless communication network, and the CPW,

2) OPTIMAL COMMUNICATION PATH MODEL IN CPW
The importance of each parameter cannot be directly reflected by the proportion coefficient due to the different units and magnitudes of the QoS parameters of the communication network. A method is proposed to balance the numerical differences between parameters to solve the problems above. The delay weight γ d , bit error rate weight γ e , bandwidth weight γ b and routing hop weight γ h are defined as shown in formula (2). The importance of each parameter can be intuitively reflected through a simple proportional coefficient.
where, D Cavg is the average delay between two nodes in CPW. E Cavg is the average bit error rate between two nodes in CPW. B Cavg is the average communication rate between two nodes in CPW. H Cmax is the maximum communication path between two nodes in CPW routing hop count. The equations (3) and (4) can be used to describe the problem of finding the optimal communication path from node v s to v d that satisfies the quality of service constraints based on the above analysis. It subjects to formula (4), while setting the minimum value of S C .
where, D max is the maximum delay of the communication path. E max is the maximum bit error rate of the communication path. B min is the minimum communication rate of the communication path. H max is the maximum number of routing hops of the communication path. The optimal communication path between the central node and all other nodes is established, and a fusion network of low-voltage power line and micro-power wireless communication with the best communication quality is formed.

3) IMPROVED ANT COLONY ALGORITHM AND ITS PATH SEARCH METHOD
An improved ant colony algorithm controlled by Brownian motion and local convergence number is proposed to VOLUME 10, 2022 solve the networking problems in cross-layer integration of low-voltage power line and micro-power wireless communication. In the ant colony algorithm with the maximum number of iterations T max , if the same result occurs in consecutive iterations of T local , the algorithm may be in a local convergence state, and T local is called the number of local convergence in this paper. The influencing factors of Brownian motion of pheromone are defined as the number of iterations and local convergence times of ant colony algorithm.
In the early stage of the algorithm, the update of ant colony pheromone is almost not affected by Brownian motion. With the increase of the number of iterations or the phenomenon of local convergence, the Brownian motion of pheromone is used to randomly adjust the pheromone distribution, which is beneficial to the algorithm to jump out of the local optimal solution and enhance its global search ability. When the number of iterations is large enough, the Brownian motion of the pheromone is weakened to speed up the convergence of the algorithm. The improved Ant colony algorithm controlled by Brownian motion and Local convergence number (ABL) combines Brownian motion and local convergence times to dynamically control the pheromone update of the path taken by the ant colony to improve the problem that the algorithm is prone to fall into local optimal solutions. ABL adopts an optimization method similar to the ant colony algorithm. It selects the optimal path and updates the pheromone iteratively, and obtains the stable path with the largest pheromone concentration, which is the optimal path. The maximum communication distance between nodes and the constraints of service quality are combined to find n candidate nodes and calculate the probability of selecting each candidate node, as shown in equations (5) and (6).
where, P(e i,j ) is the probability of node j being selected. τ (e i,j ) is the concentration of pheromone on the path ei,j. η(e i,j ) is the heuristic information on the path e i,j . Where, α and β are the influence coefficient of pheromone and the influence coefficient of heuristic information respectively. A select is the set of n nodes to be selected. It can be seen from equation (5) that the greater the pheromone concentration on the communication link and the better the communication service quality, the greater the probability of the node to be selected is selected. Taking the path cost function as the standard, the optimal path and the sub-optimal path are obtained by statistics, and the ants who walk through the optimal path release pheromone to enhance the probability of the path being selected. The pheromone update of this positive feedback mechanism is as follows (7) shown.
where, τ (v s , v d ) is a vector, which contains the pheromone of the node passed by the optimal communication path.
τ (v s , v d ) is the pheromone variable. The ρ and Q are the volatility coefficient and intensity coefficient of the pheromone, respectively. The corresponding pheromone concentration becomes larger on the communication path with better communication quality after the pheromone update. The pheromone volatilization coefficient of the algorithm is adaptively adjusted according to the number of iterations and the number of local convergence, as shown in formula (8).
where, T is the number of iterations. ρ 0 is the initial value of the pheromone volatilization coefficient. ρ min and ρ max are the minimum and maximum values of the pheromone volatilization coefficient, respectively. The pheromone volatilization coefficient is small in the early and late iteration of the algorithm,, which is conducive to the rapid convergence of the algorithm, and the pheromone volatilization coefficient is large in the middle of the iteration, which is conducive to enhancing the global search ability of the algorithm. Let the optimal path taken by the ant be R(vs, vd), and its pheromone vector be τ (vs, vd). The Brownian motion pheromone update for τ (vs, vd) is beneficial for ABL to jump out of the local optimal solution, as shown in equation (9).
where, τ (v s , v d ) is the pheromone vector before updating. δ is the step size parameter of Brownian motion. u(l) is a random number vector of length l, the elements of which conform to the standard normal distribution. The central node implements the above path search method for all other nodes, establishes the optimal communication path between each node and the central node, completes the networking process and dynamic maintenance of the CPW, that is, accomplishs low-voltage power line and micro-power wireless communication at the network layer.

C. BUSINESS LAYER INTEGRATION
The low-voltage power lines have the same MAC layer protocol as micro-power wireless communications in CPW.
The service data to be transmitted is divided into 2 subservice flows at the source node, which are transmitted in two communication networks respectively, and the received sub-service flows are combined at the destination node. The data transmission between the source node and the destination node is completed based on the fusion network. The sub-service flow distribution at the service layer of the CPW can accomplish the deep integration of the low-voltage power line and the micro-power wireless communication network, which is beneficial to accomplish the load balance between the low-voltage power line and the micro-power wireless communication. If the load difference between the two communication networks is too large, it may cause the problem of excessive data traffic and heavy network load on one of the communication networks, increasing the transmission delay. When the channel quality of a certain communication mode changes abruptly, it has a great impact on the overall communication quality of the hybrid network. Maintaining the load balance between the low-voltage power line and the micro-power wireless communication network can enhance the survivability of the hybrid communication network. The delay and communication rate of the communication link of converged network have been determined, and the allocation of sub-service flows is related to the bit error rate of the link. A service bit error rate (BER) requirement factor is proposed considering the load balancing between networks and the BER requirements of different services, which acts on the sub-service flow allocation process of the network service layer of CPW. It is assumed that each node can use power line communication and wireless communication, the service quality parameters between two adjacent nodes in the CPW are shown in formula (10).  where, C X and C Y are the proportions of sub-service flows in communication network X and Y , respectively. B X (e i,j ) and B Y (e i,j ) are the communication rates of the communication networks X and Y at e i,j , respectively. I max is the maximum value of the bit error rate demand factor of service Z . Z b is the communication rate required to transmit service Z . The BER demand factor of a service is related to the BER between CPW nodes and load balancing between networks. The corresponding BER demand factors can be set to improve user QoS requirements and load balance for services with different QoS requirements. The cross-layer fusion of lowvoltage power line and micro-power wireless communication for the information perception layer of the Internet of Things is completed, which is shown in Figure 2.

III. CPW CROSS-LAYER FUSION DESIGN
The relay nodes are selected by ABL according to the pheromone concentration of the path between nodes during the networking process, and the optimal communication path is selected according to the cost function value. It is assumed that the communication node can perceive the channel state VOLUME 10, 2022  information between the links through data transmission. The cross-layer fusion process of CPW is shown in Fig.3.

IV. SIMULATION RESULTS AND ANALYSIS A. SIMULATION CONDITIONS
In order to simulate the actual power line channel state, it is necessary to set reasonable relevant physical parameters to verify the communication fusion method proposed in this paper. Simulation experiments are carried out on the premise of the following assumptions.
(1) It can calculate the communication rate, delay, and bit error rate between it and the previous node after each node receives the data packet, and store them in the corresponding routing table.

B. PARAMETER SETTING
The simulation parameters in this paper are based on reference [30]. The time delay between any two points in the low-voltage power line and the micro-power wireless communication network is set as the compliance interval  Table 3.

C. SIMULATION ANALYSIS 1) VALIDATION OF ABL a: PERFORMANCE COMPARISON OF ABL AND ANT COLONY ALGORITHM
An optimal communication path with all other nodes are established with the central node in CPW. The average  number of iterations required for the convergence of ABL and ant colony algorithms and the average cost function value of the optimal communication path of all nodes are counted, which is shown in Table 4. It can be seen from Table 4 that ABL has better global search ability than ant colony algorithm based on the number of iterations required for ABL convergence is more than that of ant colony algorithm, and the corresponding average cost function value is lower than that of ant colony algorithm. The networking work of the central node to all other nodes can be accomplished, and the performance is better than that of ant colony algorithm.

b: PERFORMANCE COMPARISON OF ABL AND WIRELESS AD HOC NETWORK ON-DEMAND DISTANCE VECTOR ROUTING ALGORITHMS
AODV (Adhoc On-demand Distance Vector routing) algorithm is a typical routing algorithm applied to wireless ad hoc networks. The central node is set up with all other nodes to establish an optimal communication path in CPW, and the cost function value of each node's optimal communication path is obtained. The cost function value comparison of optimal communication path are shown in Figure 4.
It can be seen from Figure 4 that the cost function value of the optimal communication path found by some nodes (about 43% of total number of nodes) using ABL is lower than the cost function value of the optimal communication path found by using the AODV algorithm. The optimal communication path found by other nodes (about 57% of total number of nodes) using the two algorithms are same, so the performance of ABL is better than the AODV algorithm.

c: ABL ROUTING SEARCH VERIFICATION FOR DIFFERENT QUALITY OF SERVICE TYPES
The different networking solutions according to different service quality requirements can be provided by ABL. The voice Table 7 and Figure 6 that the optimal communication path between the central node of the dual-mode communication network and the node 70 is better than CPW in terms of delay and bit error rate. The communication rate of CPW is much higher than that of the dual-mode communication network. It has a significant impact on all nodes. The cost function value of the optimal communication path of the networking is lower than that of the dual-mode communication network, that is, its comprehensive communication service quality is better than that of the dual-mode communication network.

4) COMPARISON OF CPW AND CASCADE COMMUNICATION
The cascaded communication network can also accomplish all the access of nodes, which is compared with CPW. It is   Table 8. In the table, the cascade network 1 and the cascade network 2 represent the low-voltage power line and the micro-power wireless communication cascade communication network.

5) INFLUENCE OF BUSINESS FLOW ALLOCATION STRATEGY BASED ON BUSINESS SERVICE QUALITY REQUIREMENTS ON CPW
When node 1 communicates with node 16, considering load balancing and service bit error rate requirement factors, the sub-service flow is allocated by setting the service bit error rate requirement factor I . It is assumed that the communication rate required by the service is Z b , the relationship between E c , Z b , and I is shown in Figure 7.
When Z b = 300kbit/s, E P (e 1,16 ) = 9.017 × 10 −5 , E W (e 1,16 ) = 3.585 × 10 −5 , B P (e 1,16 ) = 198.38kbit/s, B W (e 1,16 ) = 201.70kbit/s. The maximum value of the bit error rate demand factor Imax=0.338 is known from equation (10). Table 9 shows The transmission bit error rate and inter-network load information between node 1 and node 16 is shown in table 9. There, E C is the transmission bit error rate between node 1 and node 16; Z P and Z W are the communication rates of the sub-service flows of the lowvoltage power line communication network and the micropower wireless communication network allocated by service Z b , respectively.
It can be seen from Figure 7 and Table 9 that with the gradual increase of the service bit error rate demand factor, the load gap between the low-voltage power line communication network and the micro-power wireless communication network increases, and the bit error rate of the communication link gradually decreases. The services with lower BER requirements can balance network load at the expense of increasing BER. The services with higher BER requirements can reduce BER at the expense of network load balancing. The transmission service can set the corresponding service bit error rate demand factor according to its service quality requirements, so as to comprehensively consider the service quality of the service communication and the load balance between the networks.

V. DISCUSSION
It is necessary to set reasonable relevant physical parameters to verify the communication fusion method proposed in this paper. An optimal communication path with all other nodes are established with the central node in CPW. The optimal communication path between central node and node 70 is searched in three kinds of networks respectively. When the dual-mode communication network is networked, a better communication method can be selected for information transmission according to the channel state of the low-voltage power line and the micro-power wireless communication link. The sub-service flow allocation strategy on the service layer can improve the effectiveness and reliability of CPW, and accomplish the deep integration of low-voltage power line and micro-power wireless communication network.

VI. CONCLUSION
(1) The unified MAC protocol of low-voltage power line communication and micro-power wireless communication is designed, and the integration of the two communication methods is accomplished at the MAC layer of CPW, which supports the integration of low-voltage power line and micro-power wireless communication at the network layer. (2) The ABL algorithm is proposed to complete the optimal communication path networking of CPW, and the Brownian motion related to the local convergence times is used to update the pheromone vector, which is beneficial for the algorithm to jump out of the local optimal solution. The simulation results show that the ABL algorithm has stronger global search ability than the ant colony algorithm, and it is not easy to fall into the local optimum. (3) A service bit error rate requirement factor is proposed to solve the problem of service quality requirement and load balance between networks in sub-service flow allocation. It is verified that the load balancing of the CPW and the BER of the communication link can be adjusted by bit error rate demand factor. (4) A unified communication protocol is established for the MAC layer of CPW, the optimal communication path networking is realized at the network layer, and sub-service flow distribution is performed at the service layer. The cross-layer fusion of low-voltage power line communication and micro-power wireless communication is completed. It is verified that the communication service quality of the cross-layer fusion network is better than other networks, and the communication efficiency of CPW is 398.20kbits/s.  He has been engaged in the construction, operation, and management of energy internet, and power Internet of Things for a long time. He took the lead in the overall planning, construction, and implementation of major projects, such as China Singapore Tianjin eco city smart grid comprehensive demonstration project and smart energy town, and have rich experience in the organization and implementation of demonstration projects and technology integration verification. He has rich scientific research experience and high technical level in the fields of power grid dispatching operation, safe production, and informatization.
YONG-SHAN GUO received the Bachelor of Science (Engineering) degree majoring in applied physics from the College of Physical Science and Technology, Hebei University, in 2001. He is currently working with the ICT Research Institute, State Grid Information & Telecommunication Group Company Ltd. He is mainly engaged in project research and development in power consumption and distribution and comprehensive energy services. The technologies involved include communication, control, data analysis, and related fields.
DONGDONG LV received the master's degree in control theory and control engineering from Liaoning Technical University, China, in 2018. He is currently an Engineer at State Grid Information and Communication Industry Group Company Ltd. His current research interests include power Internet of Things information communication, integrated energy applications, and power marketing technology. VOLUME 10, 2022