Improvement of wireless sensor network algorithm based on energy balance

Abstract: Hierarchical routing algorithm is an important method to extend the life of wireless sensor networks. However, most of the existing clustering algorithms have the problem of unbalanced load energy. This paper mainly focuses on the classic clustering algorithm *EACH, analyzes its basic ideas, clustering mechanism and cluster communication mode, and improves its unbalanced load energy problem, and uses MATLAB for simulation analysis. The simulation results show that the improved algorithm can balance the energy consumption of nodes, make clustering more reasonable, and effectively extend the life cycle of the network.

0 Introduction

Wireless sensor networks have been a research hotspot in the field of information technology in recent years. They integrate technologies from multiple fields such as sensors, computer science, signal and information processing, and communications. As an emerging and developing technology field, the industry is continuously deepening its research. Wireless sensor networks provide a new and effective means for humans to interact with the objective physical world. Its many characteristics make its application scope involved in military applications, industrial monitoring and control, medical monitoring, smart homes, logistics management, consumer electronics and many other fields, with broad market and industrial prospects. In August 2003, the technology review of Business Week in the United States positioned wireless sensor networks as one of the four pillar industries in the field of high technology in the 21st century.

In wireless sensor networks, energy effectiveness is an important indicator of network performance. It has very strict restrictions on energy consumption, and energy should be consumed as little as possible to achieve the purpose of extending the network life cycle. Therefore, it is very necessary to design a good routing protocol to reduce unnecessary energy consumption. This paper mainly discusses the Low Energy Adaptive Clustering Protocol (LEACH), points out the defects of the LEACH protocol, and gives corresponding solutions to optimize it.

1 Analysis of the Classic LEACH Protocol

1.1 Algorithm Description

LEAC (Low-Energy Adaptive CluSTering Hierarchy) protocol is a low-power adaptive hierarchical routing algorithm designed for wireless sensor networks. It is the earliest proposed cluster routing protocol. Its basic idea is to randomly select cluster head nodes in a cyclic manner, and other nodes are clustered according to the signal strength received from the cluster head, so that the energy load of the entire network is evenly distributed to each sensor node, thereby reducing network energy consumption and improving the overall network survival time.

The LEACH protocol defines the concept of "rounds", each round consists of the cluster establishment and stable state stages. In the cluster establishment stage, the first batch of cluster heads are selected randomly. For a node n, a random number between 0 and 1 is randomly selected for it. If this number is less than a threshold value T(n), then node n becomes the cluster head node of this round. The threshold T(n) is defined as follows:

Among them, P is the percentage of cluster head nodes in the network to the total number of nodes; r is the current round number; G is the set of nodes that have not served as cluster head nodes in the previous 1/P rounds; the symbol mod is the modulus operator.

After the cluster head node is selected, it broadcasts the information that it has become the cluster head (ADV) to the surrounding area. The non-cluster head nodes decide the cluster type they belong to based on the received signal strength. When the cluster head receives the feedback message, it allocates time slots to the nodes in the cluster (based on TDMA). In the stable stage, the nodes in the cluster send detection data to the cluster head when their time slots arrive. The cluster head node performs necessary fusion on the received data and transmits it to the base station or aggregation node. After a period of data transmission, the network re-enters the cluster establishment stage and performs the next round of cluster reconstruction, and so on.

1.2 Limitations of the LEACH Algorithm

The LEACH algorithm evenly distributes the load across the entire network, greatly saving energy loss during the communication process. The cluster head position rotation algorithm allocates the load of long-distance communication to network nodes in turn, which can extend the life of the entire system. In addition, the cluster head node uses data fusion and data compression technology when processing data, which greatly reduces the amount of data transmitted. However, the LEACH algorithm also has many shortcomings:

(1) Cluster head selection problem. The cluster heads of the LEACH protocol are randomly generated. The selection mechanism does not consider the remaining energy of the node and the number of times the node has served as a cluster head. Once a node with less remaining energy becomes a cluster head, it will quickly exhaust its energy and die prematurely. The members in its cluster will also continue to send request signals because they cannot receive information from the dead cluster head, consuming a lot of energy and causing accelerated death, which reduces the survival time of the entire network.

(2) The number of cluster heads. In the random selection mechanism of the LEACH protocol, the number of cluster heads is not controlled. Therefore, it is very likely that only one or two cluster heads will be generated in a certain round, or many cluster heads will be generated. If there are too few cluster heads, the member nodes will have to communicate with the cluster heads through a long path, and the cluster heads will also receive information from a large number of nodes and forward it to the base station. Therefore, each node is overloaded; if too many cluster heads are generated, too many nodes will communicate with the base station, reducing the utilization rate of network energy.

(3) Cluster head distribution problem. In the LEACH protocol, although the cluster heads are evenly distributed statistically, due to the randomness of cluster head generation, there may be a phenomenon where the cluster head density is high in some areas and the cluster heads are sparse in some areas.

2 Optimization of LEACH algorithm

The above-mentioned deficiencies in the LEACH algorithm lead to unbalanced load energy in wireless sensor networks. This paper mainly optimizes the LEACH protocol by improving the cluster head node election algorithm. The main goal is to prevent low-energy nodes from becoming cluster heads, control the number of cluster heads to the optimal level, and reduce the uneven distribution of cluster heads in each round. This will achieve the ultimate goal of reducing system energy consumption and extending the network life cycle.

2.1 Algorithm Improvement of Cluster Head Election Mechanism

For the improved protocol of cluster head election, the threshold is improved in the literature [6]:

Where, is the current residual energy of node n, and is the initial energy of node n.

From this analysis, we can see that since the residual energy of a node is always less than its initial energy, the improved threshold value must be smaller than the original T(n) value. Although the possibility of nodes with less residual energy becoming cluster head nodes is reduced, it also reduces the chances of being able to serve as cluster head nodes in the entire network. In response to this phenomenon, this paper takes into account two parameters: the current residual energy of the node and the current average energy of the network .

In the formula, is the current residual energy of the node, is the current average energy of the network. In this way, it is ensured that the possibility of a node being selected as a cluster head node is related to the amount of its residual energy, and that the number of cluster head nodes elected in one round is the same as the expected number.

It has been confirmed in many literatures that the number of cluster heads in the network is also an important factor affecting the network life, so this paper also incorporates the optimization scheme of the number of cluster heads into the improved protocol. The optimal number of cluster heads in this paper is determined by the method in, as shown in formula (4).

Where, is the area covered by the wireless sensor network, N is the number of nodes in the area, is the amplification factor of the signal amplifier, is the energy consumed by the circuit itself for each bit of data sent or received, and is the farthest coverage distance of the cluster head node.

2.2 Specific implementation of the improved algorithm

The algorithm is optimized and described in detail as follows.

1) In the cluster establishment phase, the cluster head is determined by all nodes autonomously, and k clusters are generated in each round. The value of k is determined by equation (4).

2) Compare the residual energy of each node with the current average energy of the network estimated in the previous round. If the residual energy is greater than the current average energy of the network, it is eligible to become a cluster head candidate node; otherwise, it can only wait for the cluster head to broadcast cluster information.

3) Nodes with energy greater than the current network average energy determine whether the random number they generate is less than the threshold value T(n) (i.e., the improved formula (3) above). If it is less than, it becomes the cluster head node; if it is greater than the threshold value, it becomes a member node and waits for the cluster head to send notification information. At this point, the cluster head election phase is completed.

4) The node that becomes the cluster head must send cluster head notification information with a certain power, but it is not broadcast to the whole network. The message only includes the ID of the cluster head node and the message identifier. After that, the cluster head will wait for the joining information of the cluster members.

5) The member node selects a cluster head node with a strong signal according to the signal strength of the received ADV message, and sends a message to it requesting to join. The message only includes the node ID and the cluster head node ID.

6) The cluster head spends a certain amount of time waiting to receive the member node's joining cluster information, then stops receiving and arranges the TDMA time slots for the nodes in the cluster to send messages according to the amount of information received. The cluster head sends the TDMA time slots to the members in the cluster at the minimum power to ensure that there will be no conflicts when the member nodes communicate with the cluster head node. In this way, a certain round of clusters in the network has been established. Figure 1 is the flow chart of the improved algorithm for the cluster establishment phase.

7) After the cluster is established, the data transmission phase begins. Each node sends the collected information in its own TDMA time slot according to the established rules. After receiving the integrated information sent by each cluster head, the base station analyzes the sensed data and reflects it on the upper-level human-computer communication interface. According to the cluster head and node ID contained in the information and the power intensity when it sends the information, the average energy of the nodes in the network when sending the next round of messages is estimated, and this information is broadcast to the network to prepare for the next round of cycles. At this point, this round ends.

Figure 1 Flowchart of the improved cluster establishment algorithm

3 Algorithm Simulation and Performance Analysis

This paper simulates the improved algorithm in MATLAB environment and evaluates the performance of the algorithm by analyzing the results.

Figure 2 Node clustering status of the improved algorithm

Figure 3 Comparison of network node lifespans of the two algorithms before and after improvement

The environment is set as follows: the total number of sensor nodes is 100, the initial energy is 0.5J, and they are distributed in a square area of ​​100 m×100 m. The base station coordinates are located at (x, y) = (50, 50). The unit energy consumption for processing data , the unit energy consumption for sending data , and the energy consumption for data fusion are 5nJ/Bit/message.

Figure 2 shows the node clustering state of the improved algorithm. Each block area in the figure represents a cluster in a certain round. In each cluster, there is a small asterisk representing the cluster head, and other small circles represent member nodes. It can be seen that the cluster heads are evenly distributed in the figure, and the number and distribution state of the member nodes governed by each cluster head are also uniform and stable.

In the same environment, the total number of nodes is changed to 200, the base station coordinates are located at (x, y) = (50, 175), and the packet length is 500. Figure 3 compares the network node lifespan of the two algorithms before and after the improvement. The horizontal axis represents the number of rounds of network work, and the vertical axis represents the number of surviving nodes. It can be seen from the figure that the node mortality rate of the improved algorithm has a certain delay compared with the original algorithm. This shows that this algorithm reduces the phenomenon of premature death of nodes due to excessive energy consumption by optimizing the cluster head selection mechanism and controlling the number of cluster heads, and greatly prolongs the life cycle of the network.

4 Conclusion

This paper proposes its own optimization scheme for several problems existing in the LEACH protocol. The new algorithm introduces the current residual energy and the current average network energy as parameters into the cluster head election mechanism, and incorporates the optimal number of cluster heads solution. In the simulation experiment, the algorithm before and after the improvement is compared and analyzed. The results show that this optimization scheme can make the node distribution more reasonable, better balance the energy consumption in the network, and extend the life cycle of the entire network to a certain extent.