Abstract
The Internet of Things grew rapidly, and many services, applications, sensor-embedded electronic devices, and related protocols were created and are still being developed. The Internet of Things (IoT) allows physically existing things to see, hear, think, and perform a significant task by allowing them to interact with one another and exchange valuable knowledge when making decisions and caring out their vital tasks. The fifth-generation (5G) communications require that the Internet of Things (IoT) is aided greatly by wireless sensor networks, which serve as a permanent layer for it. A wireless sensor network comprises a collection of sensor nodes to monitor and transmit data to the destination known as the sink. The sink (or base station) is the endpoint of data transmission in every round. The major concerns of IoT-based WSNs are improving the network lifetime and energy efficiency. In the proposed system, Optimal Cluster-Based Routing (Optimal-CBR), the energy efficiency, and network lifetime are improved using a hierarchical routing approach for applications on the IoT in the 5G environment and beyond. The Optimal-CBR protocol uses the k-means algorithm for clustering the nodes and the multihop approach for chain routing. The clustering phase is invoked until two-thirds of the nodes are dead and then the chaining phase is invoked for the rest of the data transmission. The nodes are clustered using the basic k-means algorithm during the cluster phase and the highest energy of the node nearest to the centroid is selected as the cluster head (CH). The CH collects the packets from its members and forwards them to the base station (BS). During the chaining phase, since two-thirds of the nodes are dead and the residual energy is insufficient for clustering, the remaining nodes perform multihop routing to create chaining until the data are transmitted to the BS. This enriches the energy efficiency and the network lifespan, as found in both the theoretical and simulation analyses.
1. Introduction
The Internet of Things has increased its adoption, but these are only a few of the endless fields in which it can be applied, providing infinite uses. A lot of things will be connected through the Internet of Things (IoT) and allow for automatic (human-to-machine) machine (M2M) communication. Providing all the hardware and software for the Internet of Things (IoT) with eyes and auditory and optical sensors (WSNs), the subject becomes important due to state-of-the-art applications and cutting-edge technologies.
WSN is a cornerstone of IoT, and all depend on it. WSN's main function is in the promotion and growth of IoT is allowing lower resource and life-changing services. It links tens of thousands of sensors using wireless technology. Advancement in sensors technology makes smaller, more intelligent appliances feasible for low-cost and large-scale applications. The sensor nodes are usually made up of various numbers of WSN. WSN can be used in various industrial applications for humidity, temperature, pressure, light, and movement control, as well as in agriculture, logistics, and military, and for transport and communications.
Regardless, however, existing telecommunication technologies have not yet kept up with the increasing demands of the digital age. Better performance, higher bandwidth, lower latency, and less power consumption for the Fifth Generation (5G) are required. It provides a better, more sophisticated device and a more dependable technology. Also, with all the promises of 5G, the network infrastructure coverage is a limiting factor. The 5G network system uses a millimeter band that affects the continuity of coverage. Despite increased data throughput, the 5G network has lower service availability. More subsidiary repeaters would be needed to propagate the waves in heavily populated areas, such as a megalopolis, to maintain stable data speeds. Thus, a large base station and antenna network of deployment are needed to adequately cover a 5G coverage area. Building the network and stations would not be cost-effective.
Wireless sensor networks (WSNs) fill in as a scaffold between the physical and virtual universes. These are exceptionally scattered networks of little, lightweight sensor nodes outfitted with batteries that are answerable for detecting and communicating information to the Internet. WSN is basic in giving the most challenging solution and most alluring regions for an assortment of use regions, including military observation, torrent identification, patient health monitoring, disaster surveillance and emergency management, environment checking, and mechanical computerization. Memory, processors, detecting components, batteries, and transceivers contain the sensor nodes. Accordingly, the organization of sensor nodes is deployed in the observation region, creating a huge measure of data that should be communicated to the BS. Notwithstanding, on the grounds that sensor nodes are so little, they have a few limitations as far as memory, transmission capacity, data processing, and battery life [1].
It can be contended that energy management ought to be the essential thought when designing an effective WSN. Regularly, when WSNs are used for remote area observing, a lifetime of the sensor network guarantees the efficiency of the system and reliability on data transmission. Perceiving the components that add to the energy utilization needed to support all tasks within the WSN, an enormous extent of energy is regularly utilized for data communication. By minimizing the total number of jumps and the gaps, energy utilization can be decreased. At last, as expected, the WSN's life expectancy will be adequately broadened.
Due to the immense utility potential of sensors in diverse systems, perhaps environmental monitoring, industrial automation, healthcare, target tracking, and localization, the popularity of research in wireless sensor networks (WSN) is increasing day by day. Large numbers of sensor nodes that are compact and stocked with less power are the primary components of WSNs. Sensor nodes sense, process, and transfer the observed data of the surroundings to the destination, thus making it easier to monitor the hard environments which are inconvenient to monitor otherwise. A typical WSN comprises nodes ranging from a few hundred to several thousand [2, 3]. Dynamic network topology, power constraints, heterogeneous nature of nodes, limited preloaded energy, mobility of nodes, and adaptability during node failures are the major characteristics of WSNs. The system without any routing approach disseminates the incoming packets to every link in the network through its neighbors. The transmission of the packet is guaranteed from the source to the destination since each node recognizes the data of every other node through its neighbors [4, 5]. The system does not require complex routing techniques. The main issues of flooding refer to the blindness of the resource.
For potential large-scale networks, lightweight, low-cost, and ultimately expendable sensor nodes are needed. Also, to extend the lifetime of an individual sensor node and also that of the network, each node needs to use as little power as possible because of power limitations. The node lifetime is the duration during which data can be received, transmitted, or forwarded to others by a node. Life and energy use are also critical concerns for WSNs. Routing algorithms can make intelligent decisions with a reliable lifetime estimate that can help save resources and extend the lifetime of the node.
The amalgamation of data from various sources is termed data aggregation. The sensor nodes can generate homogeneous data packets from several nodes. Aggregation of these packets reduces the number of transmissions. Partial or complete execution of the above functions can be done in each sensor node. In comparison with communication, computations consume minimum energy, thus saving an ample amount of energy. Aggregation of data is an efficient way to attain a significant saving of energy and can further lead to the traffic optimization of various routing protocols. In general, in several network architectures, highly dominant and specific nodes are allocated for aggregation and computation tasks. Much research has been done to study various algorithms and protocols to decrease the total energy consumed by the sensor network. With sensible designing of routing protocols and application layers of the operating system concerning energy conservation, the lifespan of a sensor network can be greatly increased. The algorithms and protocols should also consider the hardware, and further, they should have the ability to utilize the distinct features of transceivers and microprocessors such that the energy consumed by a sensor node gets reduced. This capability facilitates a customized solution for various sorts of sensor node designing. Various sensor networks utilize distinct sensor nodes, which further leads to collective algorithms in the field of WSNs.
Optimization of energy consumption is considered as the key objective in the study of WSN system architecture, due to the limited energy supply of each node. Clustering of nodes is done to reduce the energy consumption of the network in WSNs, by utilizing the energy efficiently and thereby improving the network lifespan. k-means clustering is one of the numerous clustering algorithms that can enhance cluster formation in WSNs. Though the k-means algorithm enhances the cluster formation, there are drawbacks due to the random selection of the initial centroid and results in the formation of an unbalanced cluster [6]. The selection of the initial centroid is enhanced in our proposed method, and the residual energy of the nodes is considered for balancing the clusters in the cluster head selection, thus resolving the creation of unbalanced clusters in the network. The optimal chain is generated based on the threshold energy of the nodes in the network and thus allows for the maximum use of the nodes that extend the lifespan of the network.
2. Related Works
In this segment, algorithms of hierarchical routing protocols are briefly discussed, and five are used for comparative purposes with the proposed system.
The protocol Low-Energy Adaptive Clustering Hierarchy (LEACH) chooses the CHs at arbitrary and sets up the cluster hierarchy. The benefit in the system refers to a centralized approach named LEACH-C (LEACH-Centralized). The starting assigns of the head of the cluster are done dependent on arbitrary likelihood and the head publicizes to the neighbor hubs to connect as a member node [7]. The member node exchanges the information with the accessible vitality to the cluster head. By utilizing time-division multiplexing access (TDMA), the nodes in the network remain in sleep mode after transmission. Here, LEACH-C is a single-hop data transmission framework, the CHs send the melded information specifically to the sink directly with the nodes’ vitality level. The node having vitality over the threshold is considered as the head for the current round, and this data is broadcast to the complete arrange. The head for the cluster is rendered for the span of that round when the hub identity matches the broadcast identity of the same hub. LEACH-C gives a break-even with a dispersion of vitality utilization between the accessible hubs. The protocol is used only for the shortest distance and it is not scalable.
A chain-based routing approach refers to a Chain-Based Hierarchical Routing Protocol (CHIRON) [8, 9] which aims to alleviate the various flaws associated with data propagation delay. In the beginning phase, the network is segregated into different fan-shaped segments. Further, the control message from the BS is delivered to every single node, and each node decides the group to which it belongs. In the second phase, the far-away node present in the BS is triggered to form a group chain within the individual group. With the aid of the greedy algorithm, the closest neighborhood node is linked to the node, which later turns into a new node that initiates in the succeeding linking step. In the third phase, as per the highest level of residual energy associated with the group nodes, the election to choose the leader node is carried out. The node located far away from the base station is initially considered to be the head of the community chain. Later, as the group chain leader, the node identified with the highest residual energy is nominated. In the fourth phase, originally, data is transmitted to the group chain head in each group through the chain. Also, together, the chain heads dispatch their collected data in a leader-by-leader transmission manner to the base station. This mechanism outputs the low-energy dissipation, and clustering overhead will occur in this environment.
The simplest clustering algorithm named k-means clustering belongs to the unsupervised clustering technique. In this clustering algorithm, the given set of nodes is partitioned into k-clusters by calculating the centroid mean value [1, 10]. Initially, the k points are randomly selected to be the centroid of k-clusters, respectively, and the nodes closest to each point are grouped to form the clusters. For the further rounds, the centroid of each cluster is calculated and the nodes closest to it are grouped to form new k-clusters. This process continues until there is no change in the clusters. The sequence of the k-means algorithm is as follows:(1)Select the k points randomly to be centroids, where k is the number of clusters(2)Assign each node to its closest centroid using Euclidean distance to form clusters(3)Compute the new centroid of every cluster(4)Repeat the second and third steps until there is no variation in the centroid of every cluster
The system supports increasing the lifetime of the network. The performance gain is maximum only when the region of the sensor is reduced.
The system employs a hierarchical clustering algorithm, namely, Energy-Aware Unequal Clustering using the Fuzzy (EAUCF) approach to extend the network lifetime [11]. In this system, cluster head selection depends on the residual energy and also the distances from the BS of the nodes. Tentative cluster heads will be elected in the network according to the random probability approach. The Fuzzy Inference System (FIS) will calculate the competition radius of the tentative cluster heads. The tentative cluster heads will collect information about residual energy from the nearby tentative cluster heads within the competition radius. If more numbers of tentative cluster heads are available within the competition radius, the lower energy nodes will be discarded from the cluster head election. The selection of the CH will not consider the node proximity which rises in the communication cost that perhaps minimizes the network’s lifespan. The system employs the balancing of energy among the clusters but has a delay in data transmission.
The Hierarchical Power-Aware Routing (HPAR) communication protocol splits the network into various zones [12]. Every single zone, which is deemed to be an entity, is a collection of location-specific sensor nodes. Therefore, the foremost step performed by HPAR is formatting all the clustered entities (zones). The following step involves the communication scheme to finalize the way information can be hierarchically directed across various zones such that the life expectancy of the network is improved. The second step can be accomplished by routing a message through a path with the highest energy over the leftover paths with the lowest energy. The respective path is known as the max-min path, and the routing scheme offers an approximation algorithm known as the max-min ZPmin algorithm, which first identifies a path associated with the lowest energy consumption with the aid of the Dijkstra algorithm. Later, it identifies another path that can raise the network’s residual energy. Further, the HPAR protocol is involved in optimizing the two solution principles. The prime benefit of this communication protocol is its attention toward the node’s transmission energy and the lowest energy present in the path. Moreover, it utilizes the zones to monitor a majority of the sensor nodes. In this system, the overhead is associated with the network while estimating the energy.
The system decreases the distance between the transit of received data but does not support a larger network.
The Position-Based Aggregator Node Election Protocol (PANEL), one of the grid-dependent hierarchical algorithms in WSN, utilizes the data of the node’s topographical location to ascertain the node’s aggregators [13, 14]. Fulfilling the requirements of both synchronous and asynchronous systems is by far the best distinguishing attribute of PANEL. In this algorithm, the network is separated into various topographical clusters. Based on the lower-left bend location of the cluster, nodes determine a reference point in every cluster. The node adjoining the reference point is selected as the cluster head. Data, in this algorithm, can be transmitted in two ways. They are the intracluster and intercluster transmission. If the data is provided to the aggregator of a distinct cluster, then it is called intracluster transmission which has the benefit of transferring the data among the cluster in the course of aggregator election. When the information is transferred between the BSs and far-away clusters, then it is termed as intercluster transmission. The system reduces the energy consumption by employing the optimal path for data transit and the balancing of load between the nodes is less.
In addition, the authors have given and analyzed an overview of the study in the energy efficiency problems and possible solutions for 5G broadband wireless access networks. Some more 5G-related work has been introduced in [15–25]. It generates an energy-saving challenge that combines storage and data transmission costs. Furthermore, strategies for 5G network resource distribution are studied to increase energy efficiency.
3. Proposed Model
The system reduces the dissipation of the individual nodes’ energy, thereby maximizing the network’s lifetime. For the selection of the cluster head, the system uses the enhanced k-means algorithm and converts it to the chain route when the threshold value is greater than the energy of the nodes in the network [26–31].
3.1. Network Model
The hubs are set arbitrarily in a locale to screen the environment ceaselessly. The total number of nodes is represented as N, where N = {n1, n2, …, nn}. The assumption has been carried out to design the Optimal-CBR protocol:(1)There should be a fixed base station above the sensor field(2)The nodes are stationary in the sensing region(3)Initially, all the nodes have an energy level(4)The cluster head performs data aggregation before forwarding the data to the next level
Figure 1 represents the topology representation of the Optimal-CBR when two-thirds of the nodes are exhausted and the remaining energy of those nodes is insufficient to form a cluster for the clustering phase. Hence, the remaining node forwards the unit of data to the BS through chain formation. The communication is based on the close ideal way approach.
3.2. Energy Model
The data transceiver energy includes the energy of the device circuitry and the volume of data transmission and reception. The vitality is required for transmission circuitry and the information parcels are transmitted. Essentially, indeed the energy is required for getting bargains with the same variables. The vitality required to transmit a unit of information is
The vitality required to get a unit of information iswhere is the vitality misfortune per bit of transceiver circuitry. Based on the transmission extend, free space (d2) or multipath (d4) propagation is used. is the enhancement vitality with distance, d.
3.3. Operation of the Optimal-CBR Algorithm
The Optimal-CBR approach uses the k-means algorithm to form clusters and chooses the CH for each cluster based on the Euclidean distance and nodes energy. The hard threshold broadcasted by the CH to the respective cluster members is the attribute value above which the node is permitted to transmit the data to the CH. Once two-thirds of the nodes are dead and the residual energy of the remaining nodes is insufficient for clustering, the nodes use the greedy approach to form a chain-like multihop routing until the BS is reached [32–42]. The Optimal-CBR algorithm is divided into two phases, namely,(1)Clustering phase(2)Chaining phase
3.3.1. Clustering Phase
In this phase, the Optimal-CBR implements the k-means algorithm in which the network nodes are arranged into k-clusters. Initially, the k nodes are arbitrarily chosen as CHs in the network. The remaining node figures the closest CH using the Euclidean distance forming k-clusters in the initial round. For the further rounds, the centroid of each cluster is calculated. The centroid of the ith node, Ci, is given aswhere “” represents the cluster member, and x and y represent the coordinates of the nodes.
The framework not only uses the distance between the centroid-based nodes but also considers the energy of the nodes for the network’s collection of CHs. Here, Ei is the nodes’ remaining energy. The selection of CHs, therefore, is the maximum of the remaining energy and the minimum distance between the nodes and is given as follows:
There is a cluster sample with a randomly selected initial CH and a centroid calculated using the centroid formula. The centroid is a virtual node at the cluster core. The node adjacent to the centroid is selected as the tentative CH. An ID is allocated to every node considering its distance from the centroid. The nodes closer to the centroid have a smaller ID than that of the nodes away from the centroid. The node comprising the next ID number is chosen as the CH if its energy is greater than the threshold. If the value is lesser than the threshold, the current CH sends the energy of the cluster member to the base station before it quits the session. BS looks at another node for the selection of CH based on the energy. If none of the nodes is above the energy of the threshold, the system forms a chain for data forwarding. The threshold energy, ETD, is calculated aswhere K is the number of clusters in the sensor region. The chosen CH broadcasts the signal, and the nodes closer to it will transmit data to the respective CHs, according to the time slot provided by the CH.
The suggested schema employs “multihop” data transfer between the nodes and the sink. On each round, the sensed data and remaining energy of every node are collected and transferred to the sink via the CHs. The entire utilization of vitality for the single cluster incorporates the utilization of vitality of the member node and the vitality utilization of the CH within the given cluster. The overall vitality devoured by the clusters is calculated aswhere is the number of member nodes, is the energy of the cluster head, and is the energy of the cluster member.
Since “N” speaks to the node count and “M” demonstrates the sensor locale, the choice of optimal clusters within the framework is given utilizing the condition
The vitality dissemination of the organization is the item of the utilization of vitality of one cluster and the total number of clusters, , displayed within the detecting locale. The overall vitality utilization by the arrange is given as
The assumption is being taken care of for the even distribution of hubs within the sensor locale. The difference from the nodes to the CH is given as
Here, ρ (r, θ) gets to be steady, when the sensors are conveyed consistently within the environment. The selection of optimal route is done usingwhere is the optimal route in a network and is the residual energy of the node.
3.3.2. Chaining Phase
When a node’s residual energy is lower (i.e., almost drained), a beacon message about the network is sent by the BS to the alive nodes. If “s” denotes the absolute number of dead nodes, then the total network nodes that are alive is given by m = N – s. When “s” is more than two-thirds (i.e., two-thirds of the nodes are dead), the BS computes the route to form a chain. In this phase, the BS creates the path through multihop chain routing, using the greedy approach. The nodes transmit the information to the BS through the chain path. The selection of the best path is done using
Data aggregating is done by the nodes in a chain path:
Here, considering a chain with node “m0” being a source, and the member i = 0, 1, …, n, the aggregation function of the node before data transmission is given as , where c corresponds to the overhead of the aggregation, while r < 1 is the compression ratio.
The process of chaining is dynamic based on the greedy approach concerning the residual energy of the live nodes in the network. An overview of the Optimal-CBR protocol is given in Figure 2.
4. Statistical Analysis
In this segment, the proposed model is compared with a few existing clustering algorithms like LEACH-C, CHIRON, and k-means. Table 1 displays the parameters of the simulation. The performance of the proposed protocol is compared with that of the existing technique, and the following metrics are considered:(1)Average energy with several rounds(2)Existence of living nodes in each round(3)First node dead comparison for validation(4)Energy dissipation of the CH(5)Packet delivery ratio(6)End-to-end delay analogy
The residual (leftover) energy of all the nodes assessed within the proposed conspire. The sum of the network’s leftover vitality of Optimal-CBR is higher when compared to that of the existing framework like LEACH-C, CHIRON, and the k-means protocol. Figure 3 shows the entirety of the leftover vitality amid each circular and the advancement accomplished by Optimal-CBR. The improvements achieved by Optimal-CBR over k-means, CHIRON, and LEACH-C are 19.24%, 34.57%, and 71.46% of energy, respectively.
The Optimal-CBR model has a higher count of alive nodes per iteration when compared with the LEACH-C, CHIRON, and k-means protocols. The count of the living nodes in each iteration is represented in Figure 4. After completion of 1200 thousand rounds, Optimal-CBR, CHIRON, k-means, and LEACH-C have 258, 173, 142, and 76 alive nodes, respectively. In k-means, CHIRON, and LEACH-C, the energy value of the nodes is less than the threshold; the nodes communicate to the base station directly, and this leads to more energy depletion. But in the Optimal-CBR protocol, the system forms a chain route through the greedy approach to transmitting the data to the base station, which consumes less energy and increases the live nodes in the network comparatively.
As the iteration progresses, the count of the alive nodes is monitored to assess the efficiency in due consideration of the network’s lifespan. As presented in Figure 5, the current model has a higher communication round than that of LEACH-C, CHIRON, and k-means. The rounds of communication for the first node dead for Optimal-CBR are 578 which is greater than the other algorithms because of the user-defined data threshold present on every node. This has reduced the energy used by the CH to aggregate the data, which, in turn, has reduced the number of times reclustering has to be done.
Figure 6 illustrates the dissipation of energy of the CHs with each algorithm. In comparison with the other protocols, the Optimal-CBR model is far more effective as it has the lowest consumption of energy with an approximate value of 0.12 J. Also, its curve is more refined than the others concerning each iteration, the reason being the short and balanced distances between the sensor nodes and the CHs.
The fraction of the number of packets collected by the sink to that transferred by the sensor nodes is termed as a packet delivery ratio. The higher the ratio, the larger the “number” of packets delivered or accumulated in the sink. From Figure 7, it is apparent that Optimal-CBR has a better packet delivery ratio than the k-means, CHIRON, and LEACH-C protocols. The chances of failure during the data delivery process are very high in the LEACH-C protocol, because of the direct transmission mode leading to the higher energy utilization of nodes. As a result, more time is drained in the process of data transfer, and a significant volume of packets is stalled.
The occurrence of a delay in the Optimal-CBR protocol is lower than that in LEACH-C, K-Mean, and CHIRON. Figure 8 compares the protocols, Optimal-CBR, LEACH-C, k-means, and CHIRON in terms of delay in network data transfer. The LEACH-C protocol has a higher delay when compared to k-means, CHIRON, and Optimal-CBR protocols. Because of the direct transmission of data, the nodes utilize a large amount of energy in the LEACH-C protocol. In LEACH-C, a particular CH is chosen, and that cluster head is permitted to carry out data collection and data delivery. The delay occurs due to the number of data transmissions to the sink and reclustering. In k-means and CHIRON, to continue the operation, the election of the CH based on neighbor node proximity, multihop data transmission, and reclustering will occur, and this process generates a delay in the sensing field. The summary of the comparison is given in Table 2.
5. Conclusion
The Optimal Cluster-Based Routing (Optimal-CBR) protocol for organizing Internet of Things-based Wireless Sensor Networks employs the k-means algorithm to construct clusters. When the residual energy of the CH is comparatively lesser than the threshold energy, a chaining phase is used to create a routing path. The cluster head is selected considering the Euclidean distance and the node’s residual energy. The outcomes of the simulation graphs signify that Optimal-CBR has lower vitality scattering within the CH, and the sum of remaining vitality devoured is moo as compared with the k-means, CHIRON, and LEACH-C protocols, thus prolonging the node’s lifespan. Hence, the current schema incorporates uniform energy distribution in all the network nodes and maximizes the transmission rounds that perhaps reduces the consumption of the energy under the 5G environment. Thus, the proposed system enhances the efficiency of the sensor nodes by minimizing the energy consumption of the nodes, which prolongs the lifetime of the network. The system does not focus on the security aspect when sharing the data through a multihop routing approach. The limitation has been overcome by employing the security mechanism in future work.
Data Availability
The data used to support the findings of this study are available from the corresponding author upon request ([email protected]).
Conflicts of Interest
The authors declare that there are no conflicts of interest.
Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) Grant funded by the Korean Government (MSIT) under Project 2021R1A2C2014333.