HYBRID SWARM OPTIMIZATION ALGORITHM FOR CLUSTERING AND ROUTING IN IOT NETWORKSDiagram
IN loT NETWORKS, INFORMATION REGARDING WSNs IS RETRIEVED.
PSO-BA SED K-MEA NS CLUSTERINGPREDICTING CLUSTER HEAD S (CHs) AND SENSOR HEADS
IMPLEMENTING THE PSO OPTIMIZATIONROUTING USING THE BE ST FITNE SS VALUEEVALUATING THE PERFoRMANCE OF THE HYBRID SWARM OPTIMIZATION ALGORITHM
Figure 1: The proposed hybrid swarm optimization algorithmic approach.
HYBRID SWARM OPTIMIZATION ALGORITHM FOR CLUSTERING AND ROUTING IN IOT NETWORKSDescription
The proposed systematic approach utilizes a hybrid swarm optimization algorithm for clustering and routing in IoT networks. The hybrid swarm optimization algorithm includes the Particle Swarm Optimization (PSO) for clustering and Glowworm Swarm Optimization (GSO) algorithm for routing in IoT networks. This proposed optimization algorithm improves the network's lifespan.
Background of the Invention:
Internet of Things (IoT) is the source platform that links numerous numbers of devices via the Internet. The Internet of Things (IoT) has a wide range of uses in daily life for ordinary citizens, government agencies, businesses, and culture as a whole. Collaboration amongst machines in the Internet of Things to put different technologies into the physical world is a difficult challenge. As per the communication concept in IoT, routing protocols are being separated into two categories namely hierarchical routing protocols and planar routing protocols. Connection networks are subdivided into various sizes of clusters. Cluster head (CH) and multiple cluster members together result from the information of every individual cluster. The cluster head (CH) is responsible for managing or controlling entire cluster member nodes that synchronize the work within the entire member nodes. The data stored in the cluster is acquired followed by data fusion processing and finally, the data is distributed among the various clusters. Sensor nodes transmit the vision data to the CH which are connected and the data from CH is forwarded to the base station or sink node, this correspondingly represents that the CH consumes higher energy at a faster rate. In WSN architecture, CH selection is a major issue. The features such as clustering process and sensor distribution mode, clustering process, changes in the clustering count, cluster head mobility, changes in cluster size, several factors, and intra-cluster communication technique are considered for selecting the appropriate cluster head (CH). The key goal of swarm optimization is to increase the WSN's transmission range. The linear system transmission approach is designed using a hybrid particle swarm that improves the performance of WSN. The PSO's pareto front filtering capabilities are enhanced when a new hybrid swarm optimization algorithm is proposed for more objective particles. Each particle is divided into multiple swarms, each of which is searching for the target region in a complex manner over a few spots on the Pareto Front Range. In current clustering protocols, a sensor node can be left independent of clusters; these nodes are referred to as residual nodes and due to the residual node the lifespan of the network is minimized. The creation of residual nodes can be entirely removed to improve the service's lifetime.
Rao et al developed a wireless sensor network with a particle swarm optimization based on an energy-efficient cluster head selection algorithm. Clustering is considered as amongst the most effective energy-saving mechanisms for Wireless Sensor Networks (WSNs). Depending on
WSN, a hierarchical cluster includes Cluster heads (CHs) that absorb higher energy because of the heavy workload in receiving and accessing data via the sensor nodes member followed by the transmission of aggregated data to the base station. As a result, the appropriate choice of CHs is critical for preserving the sensor nodes with the appropriate energy supply and extending the lifespan of WSNs. The proposed PSO-ECHS is an energy-efficient cluster head algorithm with a Particle Swarm Optimization (PSO). The model is built around an effective encoding particle and fitness function structure. They evaluated different parameters namely intra-cluster length, sensor nodes remanent energy, and sink distance to determine the performance of the proposed Particle Swarm Optimization (PSO) technique. Cluster forming includes the combination of non-cluster head sensor nodes with CHs using a synthesized weight feature. The proposed algorithmic model has been thoroughly tested on a variety of WSN cases with different numbers of sensor nodes and CHs. The obtained results are related to certain current algorithms to show that the proposed methodology is supreme.
Cao et al developed a combinational optimization for Wireless Sensor Network (WSN) based on a clustering algorithm. Since wireless sensor networks (WSN) energy nodes are finite, it cannot be replenished until it is depleted by using the common algorithmic approach namely the clustering algorithm to extend the lifespan of Wireless communication. Moreover, modem clustering methods have certain pitfalls, such as wasting excessive energy due to sharing high residual controlling informational data among nodes or failing to get a holistic viewpoint to optimize a few competing goals. This paper proposes a new Combinatorial Optimization-based Clustering Approach (COCA) for Wireless Sensor Network to solve optimization problems. COCA resolves consequences due to clustering is obtained from the combinatorial optimization viewpoint, unlike the traditional methods that treat clustering as a discrete optimization challenge. The Wireless Sensor Network-based clustering reduces the challenges because of combinatorial optimization by introducing the coding scheme for cluster heads binary particle. Depending on the link among the vector position of particle and nodes the optimization problem is rectified. In the cluster formation, the fitness function is modeled with the parameters. Eventually, the clustering is implemented using the swarm optimization method with the binary particle. COCA is compared to three certain clustering techniques in a variety of contexts. COCA significantly outperformed in comparison with other algorithms in simulations, according to the data.
Iwendi et al utilized a metaheuristic optimization approach to achieve improved energy efficiency in the IoT networks. In recent times, the Internet of Things (IoT) has gained popularity in a variety of areas, including smart cities, agriculture, weather forecasts, smart grids, environmental services, and so on. Even though the Internet of Things has enormous promise in a variety of applications but still certain places where it can be improved. They even made the latest research on reducing the energy usage of sensors in IoT networks, which will extend the network's lifespan. In the IoT network, the most significant Cluster Head (CH) is selected in this study to maximize energy consumption. The combination of Whale Optimization Algorithm (WOA) and Simulated Annealing (SA) is considered as a hybrid metaheuristic algorithm. A few efficiency parameters namely alive nodes count, temperature, cost function, load, and residual energy are utilized to choose the best CH clusters in the IoT network. The suggested method is being contrasted to several cutting-edge optimization algorithms including the Genetic Algorithm, Artificial Bee Colony algorithm, Whale Optimization Algorithm, and Adaptive Gravitational Search algorithm. The findings show that the proposed hybrid solution outperforms current methods.
Sun et al reviewed the IoT network in concern with the swarm intelligent algorithms. The Internet of Things (IoT) seeks to link almost everything about knowledge exchange and intelligent based decision-making via the technology advanced methodologies namely cellular telecommunications, machine to machine, big data analysis, and artificial intelligence. Swarm intelligence (SI) allows individuals with low to no intelligence to participate in SI activity by teamwork. The proposed systematic approach achieves global optimization by limiting the consequence because of nonlinear complexity due to its ability of parallel processing and diffusion features. They examined and described representative SI algorithm implementations in the IoT network. The primary emphasis of this study is to enable the SI-based wireless sensor network (WSN) applications as well as the overview of relevant WSN problem statements. SI-based implementations in other IoT areas, including SI in a UAV-assisted wireless network, are also discovered. Finally, prospective study opportunities and themes are sketched.
Biabani et al developed an energy-efficient evolutionary clustering technique in IoT networks for disaster management. Wireless Sensor Networks (WSNs) are vital components of the Internet of Things (IoT) network systems, providing detecting and wireless communication. Disaster prevention is considered a security task in smart cities. Raising the lifespan of WSNs is critical to ensuring device functionality. Clustering is among the routing strategies used in WSNs to improve energy efficiency. They presented an incremental clustering and routing approach effective in balancing energy utilization in nodes while taking into account disaster-area attributes. The conceptual procedure is divided into two stages. The cluster head (CH) is collected with enhanced hybrid Particle Swarm Optimization (PSO) and Harmony Search Algorithm (HSA). They build a multi-hop routing scheme depending on improved tree encoding of PSO and a new information packet layout. In terms of absolute remaining capacity, live node count, internet connectivity, and packet distribution ratio, simulation findings for catastrophe scenarios show that the proposed solution outperforms state-of-the-art methods.
Objective of the Invention:
1. The proposed Hybrid Swarm Optimization algorithmic approach includes the association of GSO and PSO to enable effective clustering and routing in IoT networks.
2. The proposed hybrid PSO-GSO method initially collects the information regarding the sensor nodes. The PSO-based K-means optimization includes the clustering and identification of other sensor nodes and cluster heads (CH). 3. The sensor nodes send the acquired data to the CH, which then sends it to the base station via other CHs. After determining the cluster heads and sensor nodes, GSO is used to optimize them. 4. The shortest route for CHs and sensor nodes moving towards the base station is identified using the GSO optimization algorithm. Routing is done based on thefitness value received from GSO.
Summary of the Invention:
People's lifestyle preferences have been transformed by the rising relevance of Internet of Things (IoT) technologies in daily life. To enable effective communication among nodes in numerous IoT-based applications require information regarding the node location and position. Wireless Sensor Network communication is a critical component of IoT which has a wide range of implementations. WSNs are supposed to be embedded into the Internet of Things, with sensor nodes automatically accessing the internet and completing their assigned tasks. WSNs are being combined with Socially Aware Networking (SAN) is preferred in many applications. Furthermore, by combining WSN and the Internet of Vehicles (IoV), a critical analysis of the natural world can be gained, allowing for the avoidance of potentially dangerous scenarios. WSNs have a wide range of possible uses in the military, research, engineering, medical care, the atmosphere, home automation systems, area control, wildfire tracking, disaster tracking, and earthquake forecasting, among other fields. Due to the restricted resources, WSN faces a variety of restrictions, including insufficient computing and processing, lack of space, memory requirements, reduced storage, and low communication capability. As a consequence, intra-WSN network traffic must be reduced. The shorter lifespan of WSN affects the exploitation of IoT networks in a wide range.
The clustering technique in IoT is used to partition sensor nodes into classes because it offers several advantages, including resource sharing, scalability, energy savings, reduced coordination overheads, and efficient management. The sensor mobile node with a characteristic of cluster head (CH) results in the formation of the cluster during the clustering phase. In particular, the task of CH enables the sensor nodes of each cluster to coordinate with the sensor node of the other cluster to transmit information to the base station (BS) or sink. Clustering systems make use of data processing techniques to reduce the amount of data gathered at CH result in the formation of significant information and the cluster heads send the combined data to the base station.
An effective Hybrid Swarm algorithm for WSNs optimization improves efficiency by comparing energy use, and network lifespan. An appropriate energy usage, lifespan, max message transfers, and reductions in dead nodes are achieved by introducing an energy-efficient clustering and routing technique. The proposed strategy of the sensor nodes manages the energy usage of the whole network. A swarm optimization algorithm is used for both clustering and routing in IoT networks. The clustering technique includes nonlinear programming whereas the routing technique includes linear programming. When the later part delay is too long, the performance suffers greatly. All current algorithms have drawbacks such as increased energy consumption, short network lifespan, low latency, etc. To address these issues, the PSO-GSO approach is enhanced, which combines GSO routing algorithms and PSO with K-means clustering for successful optimization. The PSO-GSO-based swarm optimization approach increases network life, active nodes, efficiency, and overall packet transfer while reducing dead nodes and lowering network energy usage.
Swarm optimization is an intelligent-based general random algorithm that detects aggregation and swarm activity in foraging. The data aggregation system requires each person to obey the following strategies: (a) prevent interactions with adjacent individuals; (b) compare individual velocity with adjacent one and (c) travel to the data middle. A particle in the quest space represents a possible response to every optimization challenge in particle swarm optimization. Every particle consist of a strong value is calculated by the optimized function, and every particle acquires a respective speed that correspondingly defines its position and lengths and then particles move via the solution space with following the current optimum particle.
Wireless Sensor Network (WSN) is the backbone of the IoT network in the future. WSN is a multi hop wireless network made up of a huge amount of stable or wireless sensors. WSN can capture, store, and relay information about objects found inside the network's communication range, as well as an update to users. It has drawn a lot of interest in recent decades because of its increased adaptability with the reduced cost. WSN architecture is currently focused on improving efficiency, conserving resources, and ensuring safe communication.
Detailed Description of the Invention:
The Hybrid Swarm Optimization approach increases the lifespan of WSNs in IoT networks via the effective clustering and routing technique. In the proposed hybrid swarm optimization algorithmic approach includes two optimization methods namely Glowworm Swarm Optimization (GSO) and Particle Swarm Optimization (PSO) enables effective clustering and routing protocols. Initially, the sensor nodes in the hybrid PSO-GSO system are deployed randomly before moving on to clustering and routing methods.
Clustering based on Particle Swarm Optimization Algorithm:
Particle activation and fitness value measurement for velocity and location modifications are used in the PSO-based cluster analysis. Iteration optimizes the PSO algorithm, and enhancing generations to train the model with random values and simulations for predictions. PSO integrates data locally with data collected in the quest process, where a particle changes its current position based on both the historical data and relevant data of nearby particles and iteratively explores the essential informational data. In the iteration process, each particle is modified depending on two "finest" values. One is referred to as the fitness optimal option based on an individual's best whereas the other value obtained via the particle swarm optimizer is the "highest" value that corresponds to each particle in the community system. A particle receives a majority of the population as identify important in which the optimal value is locally better. With the estimated two finest value, the particle changes the velocity, as well as location, is predicted using the following expression,
Vid UVid + Drand ()(Qid - Yid) + D 2 Rand ()(Qgd - Yid) (1)
Yid = Yid + Vid (2)
Here, the velocity of the particle 'i' is denoted as V, the particle position is represented asYid, the earlier position of the particle is denoted as Qid, the position of the particle in the population is denoted as Qgd. The random function namely rand and Rand are ranging between 0 and 1. The acceleration coefficients namely Diand D 2 are responsible for controlling the maximum size of each particle. The inertia weight is denoted as U that controls the functioning of Diand D 2 . Global discovery is considered by high weight pressures of inertia, whereas fine-tuning the present quest field is considered by lower weight pressures of inertia.
Particle initialization is the initial stage in which the particles are distributed randomly. To estimate the centroids in the clustering, the K-means clustering method is implemented. The PSO algorithm optimizes clustering centroids and the most appropriate particle are included within the PSO is acquired perfect particle for cluster head (CH). It can be determined by calculating the length among the optimized centroid and all nodes in the surrounding area. The identification of cluster head (CH) and gateway head (GH) by deployed sensor nodes. CH is identified using the fitness value derived via PSO clustering, and GH is chosen by estimating the threshold within the network. Features of Cluster Classification are described as follows,
Sensor coverage mode: There are two kinds of sensor node coverage namely random instability coverage and deterministic coverage. The sensor nodes being distributed randomly is usually referred to as random instability coverage. Deterministic coverage has been regulated by sensor spread.
Clustering method: The clustering technique is categorized into two types namely dynamic clustering and static clustering. The network is split into several clusters during a static clustering method. A cluster is formed actively nearby a CH node in the case of dynamic clustering.
Number of clusters: The number of clusters for every turn is represented by a sequence of clusters namely fixed and variable.
Size of a cluster: The maximal distance in between CH to other nodes within the cluster is referred to as the cluster size and they are fixed and variable.
Inter-cluster communication method: Relation among sensor nodes as well as communication among sensors and the CH are referred to as inter-cluster communication. Based on multi-hop communication, there are two kinds namely single-hop jumps and multi-hop jumps.
CH selection factors: When choosing CH, the following considerations must be taken into account namely residual energy, distance to a base station, distance to sink, the core of the cluster, and length coverage to other nodes.
Routing based on Glowworm Swarm Optimization Algorithm:
Glowworms (GWs) are uniformly dispersed across the network by using the GSO method, and they possess a luminance quantity namely luciferin. When the luciferin value of glowworms is greater than that of neighboring glowworms, then the GW is highly preferred. GW includes a set of complex decision-making options and consists of distance values that encloses a complex decisions radius and luciferin levels greater than themselves and GW communicates their position to its intelligent decision space. It extends the radius of the decision space. The luciferin value obtained from the previous iteration along with the objective function in the new iteration is linked together result in luciferin value. The GSO algorithm is divided into three stages namely the luciferin update stage, movement stage, and neighborhood coverage update stage. These three stages are being used to determine the fitness value to enable routing efficiency. In each iteration process, the sensor nodes lose the energy during routing. During iterations, rather than preventing transmission loss, the routing is provided between cluster heads to enable the shortest path. The data is transmitted by the node within Cluster 1 using CH1 with shortest paths are determined using the GSO-based routing technique.
In the movement process, each glowworm must select a neighboring glowworm with a higher luciferin level in comparison with its value for the movement using a probabilistic process. The probability movement of GW j's in the direction of the neighboring glowworm 'k' is represented as follows,
Diu --_' Mk(u)-Mj(u) (3) P X~u=lEOimu lu)-Mj(u)
Here, the neighborhood of GW at time u is given as k E N5(u), N5(u)= {k: dik(u) < d (u); M5(u) < Mk(u). The variable neighborhood range with the GW j at time u is denoted as d (u), at time u the Euclidian distance in between GWs j and k is denoted asdk(u). The discrete time model for GW action is expressed as follow,
yj(u +1) yj(u) + s * (IlYk(u)-Yi(u)I Yk Yj(u) (4)
Here, at time u the GW position is denoted as yj(u) and the step size s > 0. In the neighborhood coverage update stage, the GW neighborhood range is updated with the below rule, rdi(u + 1) = min {rs, maxtO, r(u) + p(nt - IN(u)|)} (5)
Here, controlling the number of neighbors is denoted as nt and the constant parameter is denoted as P. The features of routing protocol classification are described as follows,
Distributed/ Centralized: The network is centralized with a single node is in charge of discovering and maintaining routing information. Every node gathers or constructs its routing information in an effective distributed system.
Best-effort/QoS: Best attempt contracts are those that do not guarantee a certain level of quality of service for the application's execution. QoS-aware protocols are those that offer efficient quality for device routing facilities.
Single path/ Multipath routing: Routing protocols provide one or more routes to a particular location. Single path routing will locate one or more routes by choosing the efficient path for transmitting data while ignoring the others. Multipath routing is a protocol for transmitting perceptual data that identifies, preserves, and uses various routes.
Event-driven/ Query-based routing: Event-driven and query-based application programs are all built on the routing protocol system. The sensor identifies an event after initiating the data routing in the event-driven protocol. The sink node communicates the relevant query value in the query-based protocol and addresses the query initiates data routing using the sensors.
Fault-tolerant: Topology variations and packet loss are not a problem for the protocol.
Energy-aware: The protocol takes priority routing related to energy parameters such as the power consumption of route nodes.
Loop free: The protocol is referred to as no loops if such packet's route is assumed to be free of loops. Emphatic methods must be used in the protocol to search for and prevent potential loops.