Improvement of Emergency Communication Systems Using Drones in 5G and Beyond for Safety Applications

Drones are used for public safety missions because of their communication capabilities, unmanned mission, flexible deployment, and low cost. Recently, drone-assisted emergency communication systems in disasters have been developed where instead of a single large drone, flying ad hoc networks (FANETs) are proposed through clustering. Although cluster size has an impact on the proposed system's performance, no method is provided to effectively regulate cluster size. In this paper, optimum cluster size is obtained through two distinct meta-heuristic optimization algorithms - the Cuckoo Search Algorithm (CUCO) and the Particle Swarm Algorithm (PSO). Flowcharts and algorithms of CUCO and PSO are provided. A presentation of an analytical investigation based on the Markov chain model is provided. To further validate the analytical study, simulation results are presented. Simulation shows the improvement in terms of throughput and packet dropping rate (PDR).


INTRODUCTION
Unmanned aerial vehicles (UAVs), often referred to as drones, have garnered attention in recent times owing to their versatility, independence, and wide range of applicable industries [1][2][3][4].Indeed, a multitude of applications, including those in the telecommunications, military, monitoring and surveillance, search and rescue operations, medical supply delivery, and disaster management, have been seen as being significantly facilitated by drones [5][6][7][8].Over the course of the next ten years, wireless networks that support 5G and 6G are anticipated to be crucial in supporting the fast growing and pervasive use of drones in a broad range of applications [9][10][11][12][13][14].The seamless, high-definition, instantaneous, always-on connection that customers need is the aim of the 6G.6G is expected to meet the high demands of connected devices and automation systems, such as drones and driverless automobiles, in terms of energy economy, latency, data throughput, and dependability.A complete architecture with integrated terrestrial and non-terrestrial networks is part of this [15][16][17][18].Drones are essential in many different scenarios and use cases, some of which may go beyond 5G and 6G.Drones are used in a variety of applications, including remote construction, real-time surveillance, package delivery, and media production.The use of drones has significantly increased.
Rearranging the network in the event of a complex issue that arises in communication systems after natural catastrophes, such earthquakes and floods, takes a lot of time, and the most important thing is to keep people safe.In a situation like this, swift and effective search and rescue operations are desperately required.The first 72 hours of any crisis are crucial for prompt action, and effective search and rescue operations are the only way to meet this need.Nevertheless, a snag in publicity and communication will hinder their efforts.Drone-assisted emergency networks during catastrophes have been devised recently [1], where clustering is suggested to create flying ad hoc networks (FANETs).Ad hoc networks composed of many UAVs are referred to as UAVs ad hoc networks, or flying ad hoc networks, or FANETs for short.The usage of FANETs is seen in Fig. 1.A cluster head (CH) will be present in each cluster, and it will be linked to the emergency communication vehicle (ECV).Through the ECV, CH will be in charge of facilitating communication amongst cluster members (CMs) both inside and outside of the cluster.Each and every kind of communication that occurs within, outside, and between clusters enters via the CH.We demonstrated that the suggested approach performs better than current techniques in our earlier work [1].Cluster size does, however, have an impact on system performance.Performance is dependent on cluster size since factors like as collision probability, channel congestion, and packet loss are all influenced by the number of drones in the cluster.Performance drastically suffers if the cluster size is too big since there will be a lot of packet collisions when the drone count is too high.However, due to a lack of drones, or cluster members (CMs), a tiny cluster was unable to use the radio resources that were already in place.There is no technique to effectively regulate cluster size in the research [1].In this research, we use meta-heuristic techniques to optimize cluster size and hence increase system performance.
Natural intelligence, or artificial intelligence, algorithms have been widely used as search and optimization techniques in a variety of fields lately, including science, engineering, and business [17][18][19].The process of determining the best course of action under certain restrictions for a given goal or objective is known as optimization.Scientists have proposed a novel idea that has been validated via optimization.The goal of optimization is to always get the best results.The research, solution form, and acceptable tolerance are the main points of emphasis for the strongest interpretation.In an effort to address the difficulties encountered in the past, several optimization techniques have been created and used to various disciplines [20].It is usual practice to use mathematical or classical methodologies in the formulation of optimization issues.Because of these techniques' drawbacks-namely, their inelasticity and the need to identify them using mathematical functions-scientists are drawn to develop high-performance, general-purpose alternatives that draw inspiration from natural occurrences.Particle Swarm Optimization (PSO) and Cuckoo Search Algorithm (CUCO) are widely employed in optimization issues [20][21][22][23][24].
The contributions of this study is summarized as follows: drone-assisted emergency communication systems in disasters is presented where clustering is used.The performance of the proposed system is improved through two meta-heuristic algorithms -CUCO and PSO.The cluster size is optimized by algorithms.The Markov chain model is used for an analytical investigation.The analytical analyses are backed up by simulation results, which are also presented.A comparison with the previous study [1], CUCO, and PSO is presented.Simulation shows the improvement in terms of throughput and packet dropping rate (PDR).
The remainder of the paper is structured as follows: Section II outlines the meta-heuristic optimization algorithms.Section III presents the performance analysis.Simulation results are discussed in Section IV.Lastly, Section V provides the conclusions.

A. Cuckoo Search Algorithm (CUCO)
Herd intelligence is the primary focus of the CUCO algorithm.Drones are shown functioning in ad hoc networks.The goal of the CUCO algorithm is to allow drones to communicate with one another socially.The search terminates after the genetic algorithm has finished the required number of generations.Each cluster size (CS) optimizes its size based on historical data to achieve optimal performance.It is the best-positioned drones in ad hoc networks that the CUCO algorithm primarily focuses on estimating.with a randomly determined cluster size will improve their position relative to previous iterations until the objective is met.The CUCO method has been successfully used to a variety of optimization problems [24].In the first algorithm, we see the CUCO method applied to FANETs. Figure 2 shows the flowchart of CUCO algorithm.
In essence, the following phases make up the algorithm: i.An arbitrary beginning CS is used to form a starting swarm.
ii.All drone values are exchanged inside FANETs.
iii.Each drone has a local best of the current generation (pbest).There are as many best in a pack as there are drones.iv.In modern ad hoc networks, the global best (gbest) is derived from the local best.v. Following is an update to the CS.Here, Qid is position and CSid is cluster size values, while and values are random generated numbers.The value of inertial weight is w and z1, z2 are scaling factors.
vi. Steps 2, 3, 4, and 5 should be repeated until the condition for termination is met. .

B. Particle Swarm Optimization (PSO)
PSO is essentially a herd intelligence-based algorithm.FANET-operating drones have been seen.The PSO algorithm relies on drones exchanging social information with one another.The genetic algorithm's generation count determines the search strategy.Using its prior expertise, each CS modifies its position to find the optimal spot on the course.The PSO method is primarily focused on approximating the drones' positions inside the FANET to the drones that have the best FANET positions.The condition in which this approach CS occurs is arbitrary, and often the drones in the herd are in a better position than they were in their prior moves.This procedure is repeated until the goal is obtained.Numerous optimization issues have seen the effective use of the PSO algorithm [25].The PSO method for FANETs is presented in method 2. Figure 3 displays the PSO flowchart.
The following phases make up the algorithm in its entirety: i.A startup herd is created with randomly chosen starting CS.
ii.All of the FANET's drones' conformance values are computed.
iii.The local best (pbest) of the current generation is assigned to each drone.The number of drones is equal to the number of the best in the herd.iv.In the current FANET, the local bests are used to choose the global best (gbest).
Here, Qid position and id V CS values, while 1 rand and 2 rand values are numbers that are generated randomly.w is the inertial weight value and z1, z2 are scaling factors.

III. PERFORMANCE ANALYSIS
Because UAVs can move in three dimensions (3D), obstacles in their route may be easily avoided.A 3D space with movement is shown in Figure 4.For the length of the time T, it is expected that the height h will not change.(x(t), y(t), h) is the location of the UAV represented in terms of time-varying x and y coordinates, x(t) and y(t).The exact positions of the UAV are determined by launch and recovery sites, often known as pre-and post-mission locations.Let (xd, yd, h) represent the destination and (xs, ys, h) the starting point.The symbol d will represent the distance between the starting and destination points.One way to express the minimal .As a result, there exists at least one trajectory that connects the origin and the destination.
There must be one CH per cluster.There are NCL number of CHs as a consequence.The average count of CMs inside a cluster may be expressed as follows [1] 1.
At every given slot, there is a  probability that a CM will transmit a packet, which can be given as .
ϕ represents the probability of an idle channel, which may be expressed as 1 (1 ) .
 is the probability that a minimum of one CM is broadcasting on the channel during a certain slot.Since (  - 1) CMs are vying for the channel,  may be expressed as .
The probability of a transmitted packet colliding is represented by ζ.A collision will happen if any of the remaining (  -2) CMs sends a packet within the same time period.ζ may be expressed as 2 1 (1 ) .
The probability that the current packet delivery on the channel will be successful is represented by η, which may be expressed as With a Poisson process and an arrival rate of ϖa, we may compute the probability of a packet arriving, denoted by σ as where βɛ represents the estimated time of a UAV in each Markov state.βɛ can be given as (1 ) , where the length of a packet transmission with a collision is known as βC, the period of a slot is known as βslot, and the interval of a successful packet transmission is known as βS.βS and βC can be given as , , DIF ay C S del     (15) where the size of the packet is denoted by L, the length of the MAC and PHY headers is denoted by Lh, and the data transmission rate is denoted by Rd.The duration of the ACK, or βACK, and the propagation delay is denoted by βdelay.
Assume that μ is the normalized system throughput, which is determined by dividing the average length of a transmitted payload by the average slot time.For the i th cluster, the μ can be given as .

IV. SIMULATION RESULTS
This section assesses the effects of various meta-heuristic optimization strategies in FANETs.A comparative analysis is provided between the suggested system [1] and meta-heuristic optimization techniques.With MATLAB, the simulation results are achieved.Fig. 4 shows the throughput for 50 drones with a cluster size of 5.As the number of drones increases, throughput increases up to a certain point, beyond which it decreases further as more packets compete for the same channel and more collisions occur.Meta-heuristic optimization methods always have throughput that exceeds that of the proposed system [1].When there are fewer drones, the CUCO outperforms the PSO in terms of throughput.PSO algorithm throughput is greater than CUCO when the number of drones is large.Compared to the suggested system, meta-heuristic optimization techniques are more efficient.
The PDR vs the total number of drones is shown in Fig. 5.After the allotted number of attempts, a packet will be destroyed in the event that a transmission attempt is unsuccessful.CUCO has a lower PDR than PSO when there are fewer drones present.A higher drone count results in a lower PDR for PSO.The PDR of the meta-heuristic optimization techniques is consistently less than that of the suggested system.Consequently, it is evident that metaheuristic optimization techniques enhance communication efficiency by increasing throughput and lowering PDR.

V. CONCLUSION
After disaster public safety and security is the most important thing.In this study, we optimize cluster size in FANETs using meta-heuristic optimization techniques to get the best throughput performance for emergency communication systems using drones in disaster.To establish the link between the parameters, an analytical research based on Markov chains is drawn.The results of the simulation are shown.A comparison is shown between PSO and CUCO.It is clear that meta-heuristic optimization techniques improve communication reliability and efficiency.The throughput is highest for PSO when the number of drones is greater than 25.The PDR is also lowest for PSO until the number of drones is 45.It is apparent that PSO can be a better algorithm.
Drones in a herd

Table 1 .
Values of the parameters used in the simulation