Description: Centroidal voronoi partition of target regions
Edge-enabled data sink nodes and flying sensor nodes must be positioned in order to ensure the best possible sensor coverage and collision avoidance. Computer science and telecommunications are two fields that use a mathematical idea known as Voronoi tessellation, often known as Voronoi diagrams or Voronoi cells. This method divides polytopic space into polytopic regions based on how near a set of points they are. Centrodial Voronoi Tessellations (CVTs) are a special sort of voronoi division in which all points inside the voronoi cell receive optimal coverage from the center of the cell.
Consider that k quasi-stationary sensors are to be placed over a 2-D convex area W. xi ∈ R2 represents the position of sensor i, while y denotes a point in W. Let the distance function f(∥y, xi∥) denotes the performance degradation of the sensor i placed at xi of measurement at point y ∈ W. Assume, the function f is being monotonically increasing,
f(∥y, xi∥) = ∥y − xi∥2. The total sensing performance or coverage metric over W is
J(x_1,…,x_(n))= ∫_W▒█(min@i∈{1,…,n} ) f(|(|y,x_i |)|)ρ(y)dy …….(1)
The problem of optimal deployment of sensors x1, • • •, xn to provide precise distributed sensing is known as locational optimization problem which leads to a minimization problem of the cost J. Here, ρ is the scalar probability density function on W. For exploring an environment uniformly, ρ(y) can be set as Volume(W)−1.
By utilising the Voronoi cell (Vi) concept and expressing the measure element as dρ(y) = ρ(y)dy, the locational optimisation function in (1) may be reformulated as follows:
J(x_1,…,x_(n))= ∑_(i=1)^n▒∫_(V_i)▒f(|(|y,x_i |)|)dρ(y)
Therefore, problem of partitioning a convex polytopic search region W into a number of uniform cells is addressed here by using the geometric tool of Centroidal Voronoi partition.
Let Vi = V(xi) be the collection of all points y ∈ W such that, for any j ≠ i,
∥xi −y∥2 ≤ ∥xj −y∥2. This is the Voronoi region corresponding to generators xi. If two distinct cells Vi, Vj have a non-empty intersection i.e., they share borders Vi ∩ Vj ≠ ∅, i≠ j then we refer to xi as xj ’s Voronoi neighbour.
Given a region Vi and and a density function ρ(y) the mass centroid xi* of V is defined by xi* = (∫_V▒〖yρ(y)dy〗)/(∫_V▒〖ρ(y)dy〗)
In Figure 2, given k points xi, i = 1,...,k, we can define their associated Voronoi regions Vi, i = 1,...,k. Given the regions Vi, i = 1,...,k, we can determine their mass centroids xi*, i = 1,...,k. Here, we are interested in the situation where xi = xi* for i = 1,..., k i.e., the points xi that serve as generators for the Voronoi regions Vi are themselves the mass centroids of those regions. The tessellation generated is called centroidal Voronoi tessellation. Since randomly selected points in R2 typically do not correspond to the centroids of the Voronoi regions they are connected with, this circumstance is rather unique.
Sensing nodes
Weather stations, aircraft, radiosondes, dropsondes (meteorological sensor packets dropped from high-altitude platforms), satellite-based remote sensing techniques, ground-based remote sensing techniques, and aircraft are some of the methods that may be used to gather a portion of these measurements. It is noted that scientists have long employed weather stations and weather balloons to collect meteorological data for modelling. The two main disadvantages of a weather station are its fixed location and near proximity to the earth's surface. It is unable to collect information from the boundary layers. Even though weather balloons may reach considerably greater altitudes, they still have the issue of being uncontrollable and irretrievable. Drone measurements are used to close these gaps.
The sensors equipped on UAV devices collect ambient weather information and transmit the data to the quasi-stationary base station at a high altitude platform. The Internet of Things ecosystem is absolutely necessary to transmit and receive the weather data and control command in an automated drone environment. In the case of a remote environment to send and receive the crucial information and control commands, a sophisticated network model with data routing and with low latency communication is highly recommended.
Relay UAV nodes or data mules
Data mules are a set of relay UAV nodes that are used to send the localized weather label and parameters to the main base station for additional application-level control. An opportunistic message transfer protocol is used to send data. Three distinct message transfer protocols (MTP)—opportunistic flooding approaches, probabilistic flooding, and opportunistic forward methodology—are taken into consideration in the scenario that has been put out. The benefit of the opportunistic message transfer protocol allows sensor nodes to select a transmission range that is ten times smaller than the sink node's reception range within each cell. This will guarantee that sensor nodes operate for an extended period of time and reduce the cost of transmission power. Shortest path map-based mobility is considered as the model for UAV nodes over the whole considered Voronoi region for routing.

Edge enabled flying high altitude quasi-stationary nodes
A quasi-stationary base station performs as a sink node as well as high altitude sensing node deployed at the centroid of centroidal Voronoi cell. It receives the weather data collected by a set of Internet of Drone based UAVs called sensor nodes employed in each cell. It is proposed to be deployed at high altitude in troposphere where it receives the information of temperature changes with respect to the elevation. In order to provide optimal coverage to every point within the cell, it is placed onto each cell at a specific location called centroid. Each quasi-stationary base station serves as an edge-level processing control leveraging the constrained resources of flying nodes. It helps to provide real time insight of local weather and also store localized weather parameters in a buffer for future use.
5,6. Opportunistic Message Transfer Protocol (MTP)
It is necessary to effectively route the gathered localized weather forecasts and information to the main base station for additional processing and application-level services. Following message transfer approach that utilizes opportunistic routing philosophy is suggested.
In the current deployment situation, we have considered three main types of routing: opportunistic flooding, probabilistic flooding, and opportunistic forwarding schemes. In opportunistic networks, the classical flooding strategy is known as opportunistic flooding. This category includes the epidemic routing in large part. The following operations are represented
• Initialize message M in a buffer B. Assign an ID to message which is M < −I(x).
• If the contact between two nodes happens then perform the exchange of I(x) and message summary vector.
• Check the I(x) for undelivered messages.
• Finally the undelivered messages has to be transferred for the delivery
Conversely, probabilistic flooding uses history-based phenomena and transitivity to deliver the message to the nodes. It is essentially a constrained flooding methodology. In this instance, the message is transferred depending on the likelihood that the nodes would come into contact with one another, which is stated as follows:
P(X,Y)= 〖P(X,Y)〗_old+ 〖(1-P(X,Y)〗_old) × P_enc
The Penc is the encounter parameter in this case. We can think of a good chance to route the message via X to i and i to Y, towards the destination, if X meets with i more frequently. Thus, we are able to write
P(X,i)= 〖P(X,i)〗_old+ 〖(1-P(X,i)〗_old) ×P(X,i)×P(X,Y)
Forwarding technique is on the other hand restricts the number of message copy in the network. The opportunistic forwarding strategies often uses neighbourhood discovery, data transfer and storage management policy. The delivery likelihood is a crucial parameter in this case. We can consider ‘n’ as the number of nodes, every node x ∈ n keeps track of the encounter probability of its peer y ∈ n. now set f_y^x←1/(|n|- 1 ), f_y^x is the connection likelihood. If x encounters y, then〖 f〗_y^x←f_y^x+ 1. We can compute the cost as c(i,i+1,…d)= ∑_(x=i)^(d-1)▒〖(1-(f_(x+1)^x 〗) and the destination cost will be c(d)←low [ c(i,i+1,…d)].
Main base-station with intelligent computing power
The main base stations are equipped with further forecasting and application-level control. A quasi-stationary UAV node with edge capability is present in every Vornoi cell, enabling localized forecasting and prediction. While the edge-level predictive model makes use of the limited resources of flying nodes to deliver real-time insights, combining these edge predictions with models installed in the main base station (either at ground or cloud) allows for a more thorough forecasting approach. , Claims:We Claim:
1. The proposed FANET architecture ensures that the target region of interest is divided into a number of nearly uniform cells. For the purpose of gathering data, a uniform cell area results in a uniform load distribution among the sensor nodes to cover the entire target region.
2. Division of entire target region into cells also ensure collision avoidance between sensor nodes deployed in different cells.
3. Optimal sensor coverage to every point within the cell by the receiving node or sink node is ensured.
4. The sink nodes that are proposed to be placed at high altitudes are also capable of recording variations in the weather according with respect to the altitude.
5. Real time local weather prediction is ensured by edge enabled low-resource sink nodes
6. Minimization of the cost of transmission power and long-duration operation of sensor nodes is ensured by the virtue of the opportunistic message transfer framework.