Balancing Energy Fluctuations with Multi Level Trust Model for Multi Route Selection with Rank Based Route Clusters in Smart Grids

While the Smart Grid has great potential as an energy-saving solution, the communication network's interactive operation and control will result in significant energy consumption [21]. To address these vulnerabilities and to improve energy efficiency by protecting against packet loss and long latency, a trust- based multi routing protocol is proposed that avoids delay in the smart grid in case of routing issues [22]. The proposed model is more effective in large wireless networks because it employs node trust information and selects the best routes with minimum energy consumption. Despite the presence of potential threats, SGs must still complete the task of transmitting data from one node to another throughout the network. Unpredictable environmental behavior, shifting network topologies, and shaky communication can all play a role in disrupting network services.

SGs adaptability makes it useful in a wide range of settings. While different applications have varying needs for Quality of Service (QoS) parameters including throughput [23], energy efficiency [24], delay, etc., security is always a top priority. The characteristics of a program are the main factor in determining its safety [25]. However, smart networks require distinct approaches to security from those used in traditional, infrastructure-based networks. Furthermore, the features and QoS factors of each application domain are different. While multipath routing uses many paths to provide connectivity in the event of a link loss [26]. Additionally, route discovery is not always triggered when a link fails in a multipath routing system [27]. This is because, for small values of route ‘R’, the network is fault tolerant, meaning that the disruption of network services is not caused by a single failed connection [28]. The complete architectural layers of SG are shown in Figure 1.

1.png

Figure 1. Complete SG layers architecture [19]

Multiple control systems and appropriate management including demand forecasting, technical maintenance, generation and transmission planning, etc. form the basis of the smart grid's steady operation. System operators have the critical responsibility of controlling the grid frequency to guarantee the grid's efficiency, dependability, and stability on a worldwide scale. A healthy supply-and-demand relationship between power generation and load/power consumption is crucial for frequency regulation. The frequency will increase if the supply is greater than the load, and it will decrease if the reverse is true. Because of the unpredictable nature of the load and the growing fluctuations of the supply as a result of renewable energy sources' incorporation, striking this equilibrium is challenging. Extra big fluctuations could be introduced if power lines go down or if a power plant fails. Therefore, conventional operation adjusts the supply side in response to a power imbalance by adjusting it to match the load. This necessitates that smart grid possess additional generation capacity, often known as spinning reserve, which enables them to adapt the power output in response to changes in frequency. In addition, standard grids use supplemental or non-spinning reserves that are connected to quick-response generating units that may be turned on in a matter of minutes.

Although the terms are sometimes used interchangeably to refer to a risk-free system, distinguishing between trust and security is crucial to understanding the concept of trust. Trust between nodes is typically measured as an evaluation of the other nodes' dependability. As a result, it is important to employ measures beyond trust to keep networks safe. When it comes to SGs, there are a wide variety of behaviours that are considered malicious. As a result, the SGs needs a reliable method of discovering security issues [29]. Malicious nodes in SGs are able to successfully switch between states and launch attacks on the network's resources because of the nature of nodes participation in the communication. Therefore, it is crucial to maximize efficiency to choose which nodes will participate in the data transmission. Trust models are a solid method for accomplishing this goal. Models like this can aid nodes in detecting harmful activity and making the right choices [30]. The trust-based routing models in SG is shown in Figure 2.

2.png

Figure 2. SG routing model [20]

Limitations in areas like power, storage, memory, and computation have led to the discussion of how to quantify node trust [31]. Because of this, it is crucial that scalable trust models be made available that account for these constraints. In this research, a decentralized trust model in which nodes employ both direct and indirect measures of trust is considered, and where no node keeps any trust values other than those of its neighbours. Therefore, the model's reproducibility is enhanced by this method of trust and distribution measurement. Nodes cluster for a variety of reasons, including proximity to one another and energy levels [32]. Another crucial consideration is the degree of trust. To enhance SGs security, it is crucial to provide an energy-aware trust mechanism with low complexity and overhead. Bayesian analysis is generally used for establishing route in networks.  Using probability statements, Bayesian analysis provides answers to research issues involving unknown parameters. Bayesian analysis fails to disclose which node should be chosen as a routing priority. The ability to convert one's own personal set of beliefs into a formalized mathematical prior is crucial for performing Bayesian inferences. The inference procedure of the model may take some time. If there is a large amount of available data for smart grid node data, the Bayesian strategy is not worthwhile, and the regular probability approach can accomplish the work more efficiently. The proposed methodology employs the Enhanced Maximum Likelihood Estimation technique to compute the direct and indirect trust of smart grid nodes. The estimators of the shape-scale node parameters were derived using the enhanced technique. This research considers an EEMLTM-MRS-RRC in Smart Grid that maintains multiple routes by considering the trust factors. The main contributions of this research are:

•We proposed a novel trust calculation model that calculates the direct and indirect trust calculation of each node in SG using Enhanced Maximum Likelihood Estimation model. To do this, we maximize a likelihood function so that the observed data of a node is the most probable for picking that node into the routing process, given the stated statistical model. The estimators of the shape-scale family parameters were derived using the enhanced technique and compared to Bayesian estimators depending on the informational and kernel priorities.

•We designed a model for selecting network head evaluator node among the trusted nodes. The Manhattan distance model is used to calculate the distance among the trusted nodes and node that is nearer to the threshold number of nodes is considered as an evaluator node.

•The nonlinear knapsack problem is applied on the trusted nodes to select the best trusted nodes by ranking them and Trust Linked Probabilistic Controller (TLPC) is applied to select the multiple trusted shortest routes in the network for improving the network reliability during link issues.

The SG is an advanced electrical distribution network. Its advanced communication capabilities can boost system efficiency, reduce energy use, and ensure consistent service delivery, and network integration is a key component [33]. The routing protocol is a major challenge in the development of SG's communication networks because of the specifics of the deployed environment and other SG characteristics. Home area networks (HANs) and neighborhood area networks (NANs) are two types of networks that make up the smart grid's communication backbone, and they were the focus of extensive analysis of routing protocols [34]. HAN protocols can be broken down further into those that employ wireless communication. The Medium Access Control (MAC) protocols they use determine how they are built, how much they cost, and how fast they can move data [35]. Application needs, such as those for dependability, security, and quality of service, are taken into account when classifying routing protocols for NANs in addition to the underlying communication used for routing.

The AMI application's unique routing protocols have attracted a lot of attention from researchers; these protocols pose the biggest threat to the smart grid's NAN usage. The Internet of Things (IoT) has emerged in several sectors in the last several years, including transportation, healthcare, and even academia. The Internet of Things (IoT) integrates several services to achieve its goals [36]. The goal of these services is to give consumers convenient, personalized experiences through smart actions that connect their devices to the physical environment. Because of the tremendous advancements in modern technology, attacks have grown increasingly common and complex. Malicious actors sometimes exploit the diversity of the Internet of Things (IoT) to cause confusion among consumers regarding the reliability and safety of their connected devices, as well as the service they provide. Trust is thus a security concern for IoT smart services.

To identify questionable behaviour and distinguish between harmful objects, trust management strategies have been widely used in recent years. Nevertheless, these technologies still have a way to go before they can fully address complex problems, such as dealing with large amounts of data or unpredictable behaviours. Security is a top priority in SG-based communications because of the large range of linked devices. This study presents a novel trust-assuring approach that considers a broad variety of contextual parameters, including the interactions between individual nodes and their power states, in response to the challenges of SG network security. It is also advised to use a minimum hop count approach to select a route that requires the least amount of processing time. This study develops a composite routing metric that considers all of these criteria when determining which node to employ as the subsequent link in a communication chain. As a method for improving security, trust management helps protect data and user privacy. Verifying the trustworthiness of nodes before granting them permission to request help is crucial for SG security.

The distribution of fluctuations' probabilities changes, with huge tails introduced, due to the need to recover outstanding work. Consequently, even though the energy balancing model is effective in decreasing small and medium size fluctuations, the likelihood of uncommon occurrences causing big frequency variations becomes non-negligible. An unsafe unforeseen effect of small-level load balancing could cause the smart grid to go down. Afterwards, methods to prevent such unintended consequences must be considered. In this research, proper communication between smart devices, either directly or via a hub, is an easy way to minimize or eliminate excessive frequency variations caused by the necessity of recovering unfinished tasks is considered. In contrast to other efforts, the suggested method is primarily concerned with coordinating the switching of various devices in order to maintain a constant cluster power consumption through the exchange of very little data.

The node in need of assistance checks the reliability of its neighbours before passing information along to them. The fundamental issue with current trust definition approaches is that they do not easily allow for the development of metrics and evaluation frameworks. The challenges of identity administration and access control are also directly tied to the criteria of satisfaction or trust. Taking into account the multifaceted technique to evaluate the trustworthiness of SG nodes, this research provides a novel trust-aware routing framework for SG networks. To describe a node's trustworthiness in an SG network, the multi-level trust approach took into account both its communication trust and its energy trust. Additionally, multiple paths are picked based on the hop count for a given source and destination node pair. The suggested method is resilient against resource limits and security constraints since it takes into account multiple variables in choosing a forwarding node. Whereas traditional methods only addressed one of these two issues, energy or security, at a time. The proposed model architecture is shown in Figure 3.

3.png

Figure 3. Proposed model architecture

A control system that reduces the frequency stability impact of electric power on the grid could be an asset to a smart grid. For lower electric supply, the smart grid can control load. When called upon, the smart grid may regulate the amount of electricity generated by renewable sources. While auxiliary appliances aren't required for load balancing applications, power losses can happen based on factors including energy fluctuation amplitude, smoothing level, and frequency of balanced power fluctuations. To lessen the fluctuations, trusted nodes are considered and clusters are generated. Ranks are allocated to the clusters for the effective load balancing in the selected route.

There are a number of benefits to using Manhattan distance and the nonlinear knapsack problem (NKP) for smart grid route selection. While NKP is great at improving resource allocation and satisfying different constraints, Manhattan distance is great at computing efficient routes in grid-like systems. Utilizing these approaches can greatly improve smart grid solutions, which in turn improve operational efficiency and electricity distribution dependability.

Because it is both simple and computationally efficient, Manhattan distance is ideal for use in smart grid route selection. Since most urban infrastructure only allows for horizontal and vertical mobility, this metric—which measures the distance between two places in a grid-like pattern—works wonderfully with it. Smart grid systems can find the quickest routes for distributing energy by using Manhattan distance, which reduces travel time.

One area where the nonlinear knapsack problem shines is in smart grid route selection optimization for resources. To maximize energy savings while limiting operational expenses, for example, are two examples of nonlinear constraints and goals that NKP excels at handling. Because smart grids must frequently weigh opposing concerns like cost, environmental effect, and dependability, this capacity is crucial.

For smart grids to function properly, the supply and demand for electricity must be constantly tracked. Smart grids can make rapid modifications to power generation and transmission techniques to preserve energy balance with the use of energy demand forecasting models, which estimate future loads and energy supply. Smart grids face difficulties with the unpredictability and unreliability of renewable energy sources. The proposed model flow of work is shown in the Figure 4.

4.png

Figure 4. Proposed model workflow

The smart grid node's own perception in unrestricted mode is the basis for a direct trust relationship. In a wireless network, a node can interact directly with any other node and receives any traffic within its radio range regardless of how the data was originally routed. In the smart grid network, an indirect trust relies on the other node or recommender for communication. With indirect trust, one node contacts another via the network's recommender nodes. This trust model illustrates indirect trust by demonstrating how to communicate with smart grid nodes that have already been installed throughout a network. The node trust calculation algorithm calculates trust of each node by considering the internal and external parameters. The notations represented in the proposed mathematical models are indicated in Table 1.

Table 1. Notations

Algorithm Node_Trust_Calculation

{

Consider a smart grid SG that contains nodes {SGN₁,SGN₂,…..,SGN_M}. The energy γ is allocated to the smart grid.

The nodes information in the network will be gathered and processed. The nodes data will be maintained by the network authority. The node information is used to communicate with the node and also to monitor the nodes in the network. The node information processing is performed as

$N R[M]=\sum_{n=1}^M$ phyaddr $(N(n))+\frac{\tau(N(n))}{\gamma}+\alpha+T h$

A trust value will be assigned to each node in the network that processes data. Packet transmission ratio, end-to-end delay, throughput, and normalized routing overhead are some of the metrics used to determine a node's trust factor. The calculation of the trust factor will involve determining the nodes' direct and indirect trust. The direct and indirect trust of the nodes are calculated as

$\text { TransmissionRate }(\alpha)[M]=\sum_{n=1}^M \frac{\lambda(N(n))}{\omega(N(n))}+\alpha(N(n))$

$\text { LossRate }(\omega)[M]=\sum_{n=1}^M \lambda(N(n))-\delta(N(n))$

$\text { DelayLevel }(\mu)[M]=\sum_{n=1}^M \xi(\lambda(\mathrm{~N}(\mathrm{n})))-\xi(\delta(\mathrm{N}(\mathrm{n})))$

$\text { DirectTrust }(D T)[M]=\prod_{n=1}^M \frac{\delta(N(n))}{\lambda(N(n))} -\frac{\sum_{n=1}^M \beta(N(n)) * \delta(N(n))}{\lambda(N(n)) * \xi(\lambda(\mathrm{~N}(\mathrm{n}))}$

$\text { IndirectTrust(IDT)[M] }=\prod_{n=1}^M \frac{\sigma(N(n))}{M} *(\gamma(N(n))-\tau(N(n)))$

In order to determine a node's ultimate trust degree levels, the proposed method use the Enhanced Maximum Likelihood Estimation model for both direct and indirect trust calculations within the smart grid. Enhanced maximum likelihood estimation can be employed to estimate the requirements of an assumed probability distribution using some observed data from smart grid nodes. This is accomplished by maximizing a probability function such that, according to the specified statistical model, the information that is observed of a node is most likely to be used to select that node for the routing process.

The likelihood function LF(ε; M) for M node sin the smart grid contains the data to be transmitted and unknown attributes in the network is represented as ŋ=(λ, δ) and the data packets {dp₁,dp₂,….,dp_L}. The maximum likelihood estimator (ϖ)is represented as

$\frac{d x}{d y}(M)=\sum_{n=1}^M F(\eta, \lambda)+\max (D T)+\max (I D T)+\frac{\alpha}{\omega} \text { withtheinitialnodeconditionas } \mathfrak{y}(\delta)=\max (\alpha)$

The enhanced maximum likelihood estimation is performed to calculate the final trust value as

$\frac{\overline{d x}}{\overline{d y}}(M)=\sum_{n=1}^M \max \left(\frac{d x}{d y}(n)\right)+F(\mathfrak{y}, \delta)+\frac{\max (D T)}{\min (D T)}+\frac{\max (I D T)}{\min (I D T)}$

$\text { TrustFactor }(T F)[M]=\sum_{n=1}^M \max \left(\frac{\overline{d x}}{\overline{d y}}(n)\right)$

A DT and IDT based trust estimation is one that is more than threshold 0.5, where trust values typically range from 0 to 1. If the nodes trust values sum to 1, it has the highest level of trustworthiness possible. If a device's estimated trust value is below 0.5, it is seen as displaying selfish conduct or being malevolent; devices with trust ratings below 0.3 are flagged as potentially harmful. If a device's trust score is zero, it is the most malevolent device, exhibiting the worst possible behavior or generating the greatest possible number of packets, hence posing an unlimited flood threat to the network.

The energy consumption levels of the nodes with their trust values are calculated. The energy consumption of the nodes is calculated based on the node capabilities. The node behaviour will reflect the energy consumption levels. The energy consumption levels of each node are calculated as

$\text { EnergyConsumption }(\tau)[M] = \sum_{n=1}^M[\gamma(N(n))+\tau(N(n))] * \alpha(N(n))$

}

Psecudocode: EEMLTM-MRS-RRC

γ=100MW

α=100mbps

Trth=70mbps

Th=30

Counter=1

Rank=1

Rth=80

Trust TTh=75

Disth=25

Dth=10ms

NCth=50

ITh=65

Msg[M]=getData(SG_set)

For each node SGN_i  in SG_set

For i in SG_set

addr[i]=phyaddr(SGN_i)

τ=ener(SGN_i)

re=γ(SGN_i)-τ(SGN_i)

Nodereg[i]=addr[i]+τ[i]+re[i]

i=i+1

End for

For j in SG_set

α[j])=δ(j)-ω(j)+α(j)

loss[j]=λ(j)-δ(j)

For k in SG_set

IV=δ(j)/λ(j)

DT[k]=IV-

For f in range(k)

DT[f]=IV(k)-((β(k)*δ(k)/ξ(k)*λξ(k))

IK[f]=σ(k)/range(SG_set)

IDT[f]=IK(f)*(γ(k)-τ(k))

f=f+1

k=k+1

end for

j=j+1

end for

i=i+1

end for

For each node N_i in SG_set

If ((DT(N_i)>=75) && (IDT(N_i)>=65))

NodeTID=1

else

NodeTID=0

i=i+1

End for

For each node H in SG_set

if(ŋ(δ)==max (α))

MLF[H]=max(DT(H))+max(IDT(H))+(α/ω)

Tf[H]=(max(DT)/min(DT))+(max(IDT)/min(IDT))

Ener[H]=(γ(H)+τ(H))*α(H)

H=H+1

End for

For each node N_i in SG_set

If((α(N_i)>=Trth) && (ω(N_i)<0.5))&&(µ(N_i)<dth)

NC[N_i]=counter

Else

NC[N_i]=0

Counter=counter+1

i=i+1

End for

For each node N_i in SG_set

For i range (SG_set)

For j in range(i)

ManHD[N_i]=(j+1-j)/(i+1-i)

If((ManHD(N_i)<disth)&&(NC(i)>NCth))

NHEN[N_i]=NodeTID(N_i)

i=i+1

j=j+1

End for

End for

End for

For each node N_i in SG_set

For i range (SG_set)

For j in range(i)

If(ManHD(N_i)<disth)&&(NC(i)>NCth))

If(((α(N_i)>=Trth) && (ω(N_i)<0.5))&&(µ(N_i)<dth))

Rank(N_i)=Rank

If(Rank(N_i)>=Rth)

RS[i][j]=NodeTID(N_i)(i)(j)

Else

Continue

Rank=Rank+1

i=i+1

J=j+1

End for

End for

End for

End for

The proposed approach chooses a node to act as the Network Head Evaluator Node (NHEN), keeping monitoring on everything and pinpointing the safest possible path through the network. The Manhattan distance is used to determine the top dog smart grid head node.  The Manhattan distance is a standard unit of measurement that adds together the x and y separations between points. The proposed model performs ranking of trusted nodes in the SG. Nonlinear knapsack problem is used to rank trusted nodes. The knapsack problem is a combinatorial optimization problem in which one is given a set of nodes, each of which has a weight and a value, and one must choose which nodes to include in the routing process so that the sum of the weights is less than or equal to a given limit and the sum of the values is as large as possible while still only including trusted nodes.

Algorithm NHEN_Selection

{

The trust factor of each node is calculated and based on the trust factor, based on energy consumption, the nodes capabilities are calculated. The individual node capabilities are calculated as

$\begin{gathered}\text { NodeCapabilities }(N C)[M]=\sum_{n=1}^M \text { getattr }(N R(n)) \\ N C \leftarrow\left\{\begin{array}{l}N C \leftarrow \max (\alpha(N(n))+\min (\omega(N(n))+\min (\mu(N(n)) \\ N C \leftarrow 0  \text { Otherwise }\end{array}\right.\end{gathered}$

The node distance with other nodes is calculated using the manhattan distance and the node which has nearer distance to maximum nodes in the trusted node set is considered as NHEN node. The manhattandistance of nodes is calculated among 2 nodes SGN_i and SGN_j that are in the location points (SGN(X1), SGN(Y1)) and (SGN(X2), SGN(Y2)) as

$\begin{gathered}\text { ManhattanDistance }(\mathrm{f})=\sum_{n=1}^M \frac{\sum(X 2-X 1)}{\sum(Y 2-Y 1)} +\left|\operatorname{SGN}\left(X_n\right)-\operatorname{SGN}\left(X_{n+1}\right)\right|+\left|\operatorname{SGN}\left(Y_n\right)-\operatorname{SGN}\left(Y_{n+1}\right)\right|\end{gathered}$

The proposed model selects a node from the available trusted nodes a network head evaluator node. This NHEN is selected from the trusted nodes which has the highest capabilities. The NHEN node selection is performed as

$\begin{gathered}\operatorname{NHEN}[M]=\sum_{n=1}^M \frac{\max (N C(N(n)))}{M}+\min (S G N(\mathrm{f})) \\ \left\{\begin{array}{c}\mathrm{f} \leftarrow \mathrm{min}(N C(\mathrm{GN}(\mathrm{X} 1), \mathrm{SGN}(\mathrm{Y} 1)) \\ \text { continueOtherwise }\end{array}\right.\end{gathered}$

}

In this research, a Trust Linked Probabilistic Controller (TLPC) model is used to find multiple minimum-distance routes that include only trusted nodes. This research proposed an EEMLTM-MRS-RRC in Smart Grid considering all the factors that maintains multiple routes by considering the trust factors. The proposed model is divided into three modules. Initially the node t rust calculation algorithm is implemented for processing node information and the node trust calculations are performed. After trust factors are calculated, the NHEN selection is performed that monitors the entire network. Finally, the TLPC module selects the best minimal distance trusted route and energy consumption levels are calculated.

Algorithm Trust_Linked _Probabilistic_Controller

{

The distance-based route selection process is performed that is used for communication. The distance-based route selection process is performed as:

$\text { Rselection }(R S)[M]=\sum_{n=1}^M \max (N R(n)) +\min (\mathrm{f}(N(n), \mathrm{f}(N(n+1))+\min (\tau(N(n), N(n+1)))$

$R S \leftarrow\left\{\begin{array}{c}R S[\quad] \leftarrow-\max (\alpha)+\max (D T, I D T)+\min (\omega) +\min (\mathrm{f})+\operatorname{NHEN}(N(n)) \\ \text { continue } \quad \text { Otherwise }\end{array}\right.$

The selected routes based are distance is considered for final selection for multiple routes. The Nonlinear knapsack problem is used to rank the selected routes. The process of route ranking is performed using nonlinear knapsack problem with nodes count N_i (max) with a probability function Prob_f(max) is performed as

$\begin{aligned} & \operatorname{Prob}_F[M]=\prod_{n=1}^M \max \left\{\sum_{n=1}^M R S_n\right.* N_n \mid \sum_{n=1}^M[\min (\omega)+\min (\tau)] \geq T h\end{aligned}$

The nonlinear knapsack problem is represented as

$\mathrm{NKP}[\mathrm{M}]=\max \sum_{n=1}^M \operatorname{Prob}_F(\delta, \lambda) * \sum_{n=1}^M \max \left(\operatorname{Prob}_F\left({ }^{\varphi}, \sigma\right)\right.$

$\begin{gathered}\operatorname{RouteRank}\left({ }^{\varphi}\right)=\sum_{n=1}^M \frac{\operatorname{Max}(N C(n, n+1))}{M}+\frac{\max (N K P(n))}{\max (\operatorname{Prob}_F)}+\max (R S(n, n+1))+\max (D T)+\max (I D T)+\min (\tau(n, n+1))\end{gathered}$

Step-3: The TLPC model is applied on the selected routes based on ranks. The proposed model considers minimum distance routes that include only trusted nodes. The multiple route selection process is performed as:

$M R S[M]=\sum_{i=1}^M \sum_{j=1}^M \text { selecRoute }[i][j]\left(\max \left({ }^{\varphi}\right)\right)$

$M R S \leftarrow\left\{\begin{array}{c}M R S[i][j] \leftarrow \max \left({ }^{\varphi}\right)+\max (N K P(n))+ \max \left(\operatorname{Prob}_F\right)+\min (\mathrm{f}) \\ \text { continue } \quad \text { Otherwise }\end{array}\right.$

}