An effective process of VM migration with hybrid heuristic-assisted encryption technique for secured data transmission in cloud environment

Abstract

The virtualization of hardware resources like network, memory and storage are included in the core of cloud computing and are provided with the help of Virtual Machines (VM). The issues based on reliability and security reside in its acceptance in the cloud environment during the migration of VMs. VM migration highly enhanced the manageability, performance, and fault tolerance of cloud systems. Here, a set of tasks submitted by various users are arranged in the virtual cloud computing platform by using a set of VMs. Energy efficiency is effectively attained with the help of a loadbalancing strategy and it is a critical issue in the cloud environment. During the migration of VMs, providing high security is a very important task in the cloud environment. To resolve such challenges, an effective method is proposed using an optimal key-based encryption process. The main objective of this research work is to perform the VM migration and derive the multi-objective constraints with the help of hybrid heuristic improvement. The optimal VM migration is achieved by the hybrid algorithm as Improved Binary Battle Royale with Moth-flame Optimization (IBinBRMO). It can also be used to derive the multi-objective functions by some constraints like resource utilization, active servers, makespan, energy consumption, etc. After VM migration, the data transmission should take place securely between the source and destination. To secure the data, the HybridHomophorphic and Advanced Encryption Standard(HH-AES) Algorithm, where IBinBRMO optimizes the key. After optimizing the keys, the data are securely transformed along with multi-objective functions using parameters includingthe degree of modification, hiding failure rate and information preservation rate. Thus, the effectiveness is guaranteed and analyzed with other classical models. Hence, the results illustrate that the proposed work attains better performance.

Keywords

Virtual machine migration derived multi-objective constraints improved binary battle royale with moth-flame optimization secured data transmission hybrid homophorphic and advanced encryption standard

1. Introduction

In Information Technology (IT), cloud computing is the most powerful establishment in a financial institution. In small and large-scale organizations, virtual services are used to compute and store information [1]. A wide range of services are provided with the help of recently used cloud systems, and it provides several benefits in terms of avoiding additional server requirements and reducing the utilization of power resources [2]. In the cloud computing platform, commercial virtualization applications migrate VMs to allocate complimentary resources [3]. The overload of the networks is handled very effectively while allocating the complementary resources, and these resource values cannot be extended to the appropriate limit without violating the Quality of Service (QoS) and service level agreement [4]. Cloud services support the migration of VMs in order to accomplish energy management, less hardware maintenance, server consolidation, and traffic management [5]. Initially, the migration of VMs is done manually. But, this tends to increase the cost, poor choices, and downtime due to the migration process [6]. The migration of VM is done in cloud services to decrease the cost requirements and energy consumption [7]. In order to provide adaptable and precise benefits in the application environment, automated migration approaches are introduced [8].

The VM migration process is broadly classified into two types such as non-live migration and migration [9]. The current process has been held by the non-live migration approaches, and the copy of the entire source to destination pages is resumed when the migration process is completed [10]. In the live migration approach, the current process has not stopped, and it needs less downtime for transferring the VM from the source to the target [11]. Small-scale operations are only suitable for live VM migration, but transferring entire machines to all the processes is difficult and quite complex [12]. For single live migrations, the requirements related to QoS are not considered fully, and hence, the non-live migration is selected [13]. Instead of giving restricted services to the appropriate user, the downtime is compared with the live migration, and the non-live migration is preferred by the service providers with the utilization of less downtime [14].

Many of the VM migration algorithms are developed to identify the status of the utilization of VMs, and hence, the under-utilized VMs are moved to the destination systems while increasing the resource requirements [15]. Here, the selection of VMs is important to migrating it from source to destination otherwise, the performance of the systems is affected in terms of QoS [16]. By using the live migration methods, the VMs are migrated transparently and seamlessly from one machine to another physical machine. Load balancing provides significant benefits with reduced power consumption and expenses to migrate the VMs from overloaded servers to light-loaded servers [17]. Moreover, online maintenance and proactive fault tolerance are the significant factors to provide major benefits over the migration of VMs. The main bottleneck during VM migration is memory mitigation that occurs due to the adequate dirtying of VM memory pages. Hence, several optimization algorithms are developed to analyze the total movement time, energy consumption, and the utilization of memory space. Most of the VM placement and VM migration approaches only concentrate on energy utilization and memory requirements. But, the security of data packets using transmission is not provided in VM migration approaches. Furthermore, trust, authenticity, privacy preservation, as well as integrity are not considered in cloud-based systems. Hence, to provide better VM migration with higher security, a heuristic-adopted secure mechanism is developed in this research work.

The contribution of the designed security-aware VM placement and VM migration are given below.

•
To develop a new VM placement and migration approach with an advanced hybrid encryption algorithm to provide highly secured data transmission with reduced usage of memory and power requirements.
•
To design an efficient VM placement model with the help of developedIBinBRMO by optimally selecting the number of VMs to minimize the consumption of network resources and the usage of active servers.
•
To propose an effective VM migration scheme with the usage of the suggested IBinBRMO, where the number of servers and the VMs are optimally selected to decrease the consumption of energy and improve the makespan of the cloud system.
•
To implement an IBinBRMO in the developed VM placement and VM migrations scheme to improve the performance by optimally selecting the number of VMs and servers, and the security of data transmission is attained with the help of optimal key generation by using this IBinBRMO.
•
To propose a hybrid advanced encryption scheme HH-AES with optimal key generation using IBinBRMO to improve security in terms of hiding failure rate, degree of modification, and information preservation rate.
•
The performance of the developed IBinBRMO- HH-AES-based VM placement and migration model is ensured through distinct heuristic algorithms concerning several objective functions.

The remaining sections used in the developed IBinBRMO- HH-AES-based VM placement and migration model are described as follows. The features and disadvantages of conventional VM placement and migration approaches are illustrated in Section 2. The description of the dataset and the developed IBinBRMO are enumerated in Section 3. The VM placement and VM migration with its objective function are summarized in Section 4. The proposed HH-AES-based encryption scheme is illustrated in Section 5. The comparative analysis is explained in Section 6, and the conclusion of the proposed model is given in Section 7.
2. Literature survey

2.1 Related works

In 2022, Saxena et al. [18] have suggested a multi-objective-based VM placement approach in order to guarantee the energy-efficient distribution of physical resources with decreased intercommunication delay. This has emphasized timely execution and provided higher security over the transmitted data. Here, a Non-dominated genetic algorithm with whale optimization has been inspired to provide higher effectiveness over the suggested model. The effectiveness validation was made by comparing the dynamic and static approaches, but the suggested model attained less power consumption, intercommunication cost, and execution time than others.

In 2020, Inokuchi and Kourai [19] have demonstrated a VM management approach based on strong user bindings in semi-trusted cloud environments. The user bindings were strongly provided with the help of UVBond. In the trusted hypervisor, the UBBond booted the VM user by performing decryption over the encrypted disk. Here, hypercell automata have been utilized to provide a higher-level semantic gap between the low-level hypercells and the high-level management commands. By using this approach, the overhead of the network has been highly minimized.

In 2013, Li et al. [20] have introduced a Cyberlive approach for providing flexible collaboration for enabling desktop application sharing in a virtual environment on the basis of the virtualization infrastructure. Here, the view privileges were distinguished with the help of a secure access mechanism, and the events were tracked with the usage of a window filtering mechanism for delivering desktops for various users. The performance of the VM management systems was very high, and it was highly effective and useful.

In 2019, Pal et al. [21] have suggested a probabilistic resource migration algorithm for migrating VMs in the cross-cloud and public cloud. Here, the data has been privately and efficiently owned by third-party servers and cloud owners. This method was highly reliable and efficient. The important instances to be considered when sharing a huge volume of data were utilized using the memory, time and CPU. This method has also computed the VM memory flow of total memory during the movement of VMs.

In 2011, Danev et al. [22] have enabled secure VM-virtual Trusted Platform Module placement in private clouds. The classical platform processes should not impede the TPM virtualization. This method has provided strong security when compared to the previous solutions, and the feasibility of such systems was high when compared to other models. Here, the hypervisor was directly integrated by downloading the open-source software. The performance was analyzed to discuss the effectiveness of the developed model.

In 2021, Doyle et al. [23] have introduced a new lightweight framework named as blockchainBus for the secure migration of virtual machines using blockchain in cloud federation networks. This framework has been employed in the cloud environment of Microsoft Azure using the HyperLedger solution. After implementing the environment, the overhead of the network has been determined. The performance in regard to VM migration time has been considered for validating the efficiency of the BlockchainBus model. The obtained migration time was very shorter than the total movement time, and hence, this result demonstrated that the performance was higher in the BlockchainBus framework.

In 2019, Alkadi et al. [24] have introduced an anomaly detection framework for visualizing outsider attacks and insider attacks from the cloud environment. Here, the Gaussian mixture models and outlier localization-based mixture models have been utilized to fit the network, and the abnormal sequences in the network traffic were discovered with the help of the outlier factor function. The performance was analyzed concerning different anomaly detection approaches, and the suggested model provided a better outcome in detecting anomalies.

In 2020, Yu et al. [25] have designed a VM movement and management approach based on a physical server named a trust verification server for giving trust services based on the specification of the trust platform module. Here, the noise and interferences were effectively dealt with via non-interference trust measurement approach. This suggested model could solve complex cloud security issues very efficiently.

In 2020, Eirini et al. [31] have developed an effective algorithm for balancing the load in the 5G-VCC system. Here, the Modified Ant Colony Optimization (MACO) algorithm was adapted and has also been inspired by the ant’s natural behaviour. Moreover, the weight value has been assigned for each 5G-VCC. In the optimal route-selecting phase, the highest weight was selected with the help of the MACO algorithm. The result analysis has shown that the developed model has attained enhanced performance in the 5G-VCC systems.

In 2021, Eirini et al. [32] have implemented a novel network slicing scheme for tuning the performance of 5G vehicular networks. However, the LTE vehicle-to-everything (LTE-V2X) technology was implemented using the Road Side Unit (RSU) and Aerial Relay Nodes (ARN). While, the position of each RSU has been tuned using the fuzzy Multi-Attribute Decision-Making (fuzzy MADM) model. Often, the QoS and SINR factors have been monitored using the developed model. The performance has been validated regarding throughput, packet transfer delay, jitter and packet loss ratio using the developed method.

Table 1
Features and disadvantages of existing VM Placement and VM Migration approaches

Author [citation]	Methodology	Features	Challenges
Saxena et al. [18]	SM-VMP	• It highly decreases the intercommunication delay. • It provides high energy as well as resources distribution of physical resources.	• It does not provide proper security to the transmitted data. • It has a poor selection of VMs to be migrated from the source to the destination, and hence, the effectiveness is slightly reduced.
Inokuchi and Kourai [19]	AES-NI	• It provides negligible overhead in remote VM management. • It effectively prevents the data from inside attackers.	• It does not prevent the lago attack. • The external commands are invoked by several management commands.
Li et al. [20]	CyberliveApp	• It computes the hidden areas of windows by tracking the window operation events in real time. • It provides live application sharing and migration between VMs.	• It does not provide flexible collaborations for cloud-based computing environments. • It requires virtualization-based software platforms to provide effective performance.
Pal et al. [21]	PRMA	• It achieves VM migration with less downtime. • It effectively reduces the workload during migration.	• It is inapplicable to migrate between various subnet IPs. • It does not provide proper security over the transmission medium.
Danev et al. [22]	TPM module	• It provides stronger security over the transmitted data. • It provides greater feasibility.	• It does not have the possibility to secure live VM-vTPM migration. • It does not support realistic virtualized environments.
Doyle et al. [23]	Blockchain	• It improves the performance and resilience of VM migration. • It highly prevents vendor lock-in.	• It needs to reduce the system overhead. • It does not consider the system’s robustness.
Alkadi et al. [24]	MLO	• It provides higher elasticity, tolerance, flexibility, and efficiency. • It effectively determines the attack records from normal ones.	• It does not calculate the service level agreement evaluation metrics. • It does not support real cloud systems.
Yu et al. [25]	TPM	• It effectively guarantees its own security. • It effectively solves complex cloud security issues.	• It does not have the ability to ensure reliable network transmissions via cloud platforms. • It does not provide the most trust features for many specific services.

In 2021, Emmanouil et al. [33] have suggested a slicing scheme for 5G-VCC systems to enhance the performance in modern vehicular services. Here, the estimation of the user satisfactory grade was considered to evaluate the performance. The higher satisfactory grade has shown to allocate the user service with the help of Resource Blocks (RBs) and Point of Access (PoA). Moreover, the lower satisfactory grade has shown the Virtual Resource Pool (VRP), which was located at the Software Defined Network (SDN) and also it has been controlled by the required services. Thus, the outcome of the suggested model has been validated in terms of packet transfer delay, jitter, and throughput.

2.2 Problem statement

The secure allocation of VMs is important to facilitate elastic and cost-effective computing benefits for cloud users. The inefficient sharing of VM leads to increases the intercommunication costs, resource wastage, security breaches, and excessive power consumption. In order to provide higher security over the stored data concerning timely execution is performed with the help of several protocols and algorithms. The recently developed VM migration approaches with their features and demerits are illustrated in Table 1. SM-VMP [18] highly decreases the intercommunication delay. In addition, it provides a highly energy-efficient distribution of physical resources. But, it does not provide proper security for the transmitted data. Furthermore, it has a poor selection of VMs to be migrated from the source to the destination, and hence, the effectiveness is slightly reduced. AES-NI [19] provides negligible overhead in remote VM management. On the other hand, it effectively prevents the data from inside attackers. Yet, it does not prevent the lago attack. Moreover, the external commands are invoked by several management commands. CyberliveApp [20] computes the hidden areas of windows by tracking the window operation events in real-time. Furthermore, it provides live application sharing and migration between VMs. However, it does not provide flexible collaborations for cloud-based computing environments. In addition, it requires virtualization-based software platforms to provide effective performance. PRMA [21] achieves VM migration with less downtime. In addition, it effectively reduces the workload during migration. But, it is inapplicable to migrate between various subnet IPs. Moreover, it does not provide proper security over the transmission medium. TPM module [22] provides stronger security over the transmitted data. On the other hand, it provides greater feasibility. However, it does not have the possibility to secure live VM-vTPM migration. For instance, it does not support realistic virtualized environments. Blockchain [23] improves the performance and resilience of VM migration. Furthermore, it highly prevents vendor lock-in. But, it needs to reduce the system overhead. Nonetheless, it does not consider the system’s robustness. MLO [24] needs to reduce the system overhead and also, it does not consider the system’s robustness. Yet, it does not calculate the service level agreement evaluation metrics. Moreover, it does not support real cloud systems. TPM [25] effectively guarantees its own security. And also effectively solves complex cloud security issues. But, it does not have the ability to ensure reliable network transmissions via the cloud platforms. Furthermore, it does not provide the most trust features for many specific services. To address these issues, optimal VM placement and VM migration approaches are developed with encryption standards to provide higher security.

3. Block-wise schematic representation of vm migration with secure data transmission in the cloud environment

3.1 Proposed VM migration and VM placement in cloud environment

Virtualization is mainly adopted in mobile computing, resource sharing, portability requirements, and independent platforms to maximize utilization. During migration, availability, load balancing, and power management are the significant factors to be considered. To provide residual dependencies among the host machines are a significant issue in conventional VM migration schemes. The utilization of memory and resource requirements are needed to be highly minimized in these VM migration approaches. Moreover, the system overload and the cost requirements are high in the existing VM migration approaches. Live migration schemes solve these issues related to maintenance, overloading, load balancing, utilization of resources and time, and computational complexities. But, the live migration techniques do not focus on security because it is susceptible to several attacks like denial of service, man-in-the-middle attack, and stack overflow attack. During VM migration, the data can be tampered or sniffed, and hence, encryption algorithms are introduced to provide higher security during data transmission. But, analyzing the security threats in the network is slightly different. This has been accomplished by the newly developed optimization strategy-based VM placement and VM migration with a hybrid encryption scheme, and the architectural representation of the system is given in Fig. 1.

A new VM placement and VM migration approach with a hybrid encryption standard is introduced to perform the VM placement and migration very effectively with the help of a heuristic algorithm to reduce the utilization of memory resources, computational time, and overload with high security over the transmitted data. The required data are garnered from various online links. Initially, the developed IBinBRMO algorithm is used to optimally select the number of VMs to perform VM placement to decrease the number of active servers and the consumption of resources. Then, the same IBinBRMO algorithm is used for the migration of VMs in order to decrease the power utilization and makespan. After, the hybrid encryption scheme named HH-AES is introduced to provide higher security over the transmitted data, where the objectives related to high data hiding rate and high data modification rate are achieved with the usage of developed IBinBRMO. Finally, the efficacy of the suggested model is compared over various optimization strategies to examine the effectiveness of the suggested VM placement and VM migration with a hybrid encryption scheme.

Figure 1.

Structural illustration of the developed Security aware VM Placement and VM Migration approach.

3.2 Developed IBinBRMO

The implemented IBinBRMO is used in the developed security-aware VM placement and VM migration approach to optimally select the VMs for the placement to reduce the utilization of CPU resources and active server, to decrease the energy utilization and makespan by optimally selecting the number of VMs to improve the performance over VM migration. Moreover, the optimal key is generated with the usage of the proposed IBinBRMO for providing higher security over data transmission. The BinBRO algorithm is used in the approach because it converges faster than the other algorithms, and the MOA provides a better balancing ability between exploitation and exploration. But, these algorithms are struggled to solve local optima and global optima problems. Hence, to solve the global optima problems and improve the generalization effects, the IBinBRMO algorithm is developed. In this suggested IBinBRMO, the position of the candidate solutions is upgraded using the newly created concept, which is given in Eq. (1).

$\displaystyle l=\left({b-1}\right)*\delta+1$ (1)

Here, the term $b$ represents the variable that linearly decreases from $\left[{0,1}\right]$ , and the parameter $\delta$ is the randomly distributed number that is determined using Eq. (2).

$\displaystyle\delta=\left(\left.\left({\left({u*\frac{pl_{\textit{Nr}}}{% \textit{IN}_{\textit{MX}}}}\right)*\sqrt{u}}\right)\right/\textit{pl}_{\textit% {Nr}}\right)$ (2)

Here, the term $u$ is the random attribute selected in the range of $\left[{0,1}\right]$ , $\textit{pl}_{\textit{Nr}}$ is the population count and $\textit{IN}_{\textit{MX}}$ is the maximum number of iterations. If the obtained value is $l>$ 1, then update the position using MOA otherwise, update the position using BinBRO. While adopting this newly modified random parameter concept in the conventional algorithm provides a better solution over VM placement, VM migration and optimal key generation, and hence, the effectiveness of the suggested security-aware VM placement and VM migration approach is greatly improved.

MOA: The MOA algorithm is initialized by assigning the problem variables as the moths’ position in the predefined solution space, and the population of moths is assumed as the candidate solution. In the problem space, the moths are flying either in a hyperdimensional plane of 1D, 2D and 3D space. The number of problem variables is indicated by $v$ , and the number of moths is indicated by $s$ . The return value of the objective is assumed as the fitness value, and the finally obtained moth fitness function is assumed to be the best fitness function. The flames and moths are the solutions in the search space, and each should be updated for all the iterations. The moths are moving around the solution space, and then they are assumed as the actual search agents, and the flames are dropped by the search agents, then they should be considered as the pins or flags in the problem space. Therefore, the flames are surrounded by the moths to get the best solution. The global approximate of the optimization problems is given in Eq. (3).

$\displaystyle\textit{Mf}=\left({\textit{P},\textit{F},\textit{N}}\right)$ (3)

Here, the random population of the method is denoted by the term $P$ , which is selected based on the fitness function. It is mathematically defined in Eq. (4).

$\displaystyle\textit{P}:\phi\to\left\{{\textit{{MT}, {OMT}}}\right\}$ (4)

The term $F$ is the representation of the movement of moths search around the solution space, and it is considered to be the main function. Finally, a matrix MT was received based on the main function, and the updated value was returned successfully that, as denoted in Eq. (5).

$\displaystyle\textit{F}:\textit{MT}\to\textit{MT}$ (5)

If the termination criteria are met if the $N$ value returns its true value, if it is false, then assume that the termination criteria are not fulfilled.

$\displaystyle\textit{N}:\textit{MT}\to\left\{{\textit{TRUE},\textit{FALSE}}\right\}$ (6)

Moreover, the upper and the lower bound attributes also initialized in the solution space given in Eqs (7) and (8), respectively.

$\displaystyle\textit{upB}=\left[{\textit{upB}_{1},\textit{upB}_{2},\textit{upB% }_{3},...,\textit{upB}_{s-1},\textit{upB}_{s}}\right]$ (7)

The upper bound value for $v^{\text{th}}$ variable is denoted by $\textit{upB}_{v}$ .

$\displaystyle\textit{lwB}=\left[{\textit{lwB}_{1},\textit{lwB}_{2},\textit{lwB% }_{3},...,\textit{lwB}_{s-1},\textit{lwB}_{s}}\right]$ (8)

The lower bound value for $v^{\text{th}}$ variable is denoted by $\textit{lwB}_{v}$ .

After, initializing all the parameters, the function $F$ is run if it returns the termination condition value. In this MFO, the inspiration is the transverse orientation, where the moths move around the problem space. Here, the position is upgraded, and it is mathematically expressed in Eq. (9).

$\displaystyle\textit{MT}_{v}=\textit{FS}\left({\textit{MT}_{v},\textit{Le}_{h}% }\right)$ (9)

Here, the $h^{\text{th}}$ flame or pin is denoted by $\textit{Le}_{h}$ , the $v^{\text{th}}$ moth is indicated by $\textit{MT}_{v}$ , and the spiral function of MFO is denoted by FS. The spiral function can be calculated based on the condition of the initial point that is started from the moth; the final point denotes the position of flames, and the fluctuation should not exceed the value of search space.

The logarithmic function of spiral updating behavior is given in Eq. (10).

$\displaystyle\textit{FS}\left({\textit{MT}_{v},\textit{Le}_{h}}\right)=DC_{v}.% \exp^{\textit{gl}}.\cos\left({2\pi l}\right)+\textit{Le}_{h}$ (10)

The distance parameter of the $v^{\text{th}}$ moth for $h^{\text{th}}$ the flame is denoted by $\textit{DC}_{v}$ , the shape of the logarithmic spiral is denoted by gl, where $l$ is the randomly distributed attribute that lies in $\left[{-1,1}\right]$ . The distance metric $\textit{DC}_{v}$ is estimated using Eq. (11).

$\displaystyle\textit{DC}_{v}=\left|{\textit{Le}_{h}-\textit{MT}_{v}}\right|$ (11)

Based on the value of $l$ , the position of the neighbourhood is changed, and also the frequency of the position is also updated to guarantee the exploitation of the flames.

BinBRO: It is a population-based approach, where all the solutions are known as the soldier. The soldier tries to defeat the adjacent neighbour. To defeat the nearest neighbours, the soldier is placed in the best position. The BRO is initiated with a random population, and that is uniformly spread throughout the solution space. The soldiers defeat the nearest neighbours by hurting others and damaging others by finding a better position. The damage level of the soldiers at $z^{\text{th}}$ level over the population is determined based on $\textit{z}_{\textit{a .dm}}=\textit{z}_{\textit{a .dm}}+1$ . The soldiers experienced the damage, and then they immediately alter their position. The damaged soldier’s attack the opponents from another side, and in the exploitation phase, the soldiers move towards a damaged position to a better position, and it is estimated based on Eq. (12).

$\displaystyle z_{\textit{dm,a}}=z_{\textit{dm,a}}+\delta\left({z_{\textit{Bst,% a}}-z_{\textit{dm,a}}}\right)$ (12)

The random attribute $\delta$ is distributed uniformly in the range of $\left[{0,1}\right]$ , and it damages the soldier that is seen in the direction of $a$ . In the next iteration, the damaged soldiers hurt their nearest neighbors and then $\textit{z}_{\textit{a .dm}}$ are set to $0$ . The behavior of the soldier comes backward to the problem space, are indicated in Eq. (13).

$\displaystyle z_{\textit{dm,a}}=\delta\left({\textit{upB}_{a}-\textit{lwB}_{a}% }\right)\textit{lwB}_{a}$ (13)

The upper bounds and the lower bounds in the problem dimension space are denoted by the terms $\textit{upB}_{a}$ and $\textit{{lwB}}_{a}$ , respectively. The problems begin to shrink downward to a better solution. The initial value was assigned to $\vartheta=\log_{10}\left({Cir_{\textit{Max}}}\right)$ , where $Cir_{\textit{Max}}$ is the maximum iteration number that is given in Eq. (14).

$\displaystyle\vartheta=\vartheta+\textit{Rnd}\left({\frac{\vartheta}{2}}\right)$ (14)

The upper and lower bound values are given in Eq. (15) and Eq. (16), respectively.

$\displaystyle\textit{lwB}_{a}=z_{\textit{Bst,a}}-\sigma\left({\overline{z_{a}}% }\right)$ (15) $\displaystyle\textit{upB}_{a}=z_{\textit{Bst,a}}-\sigma\left({\overline{z_{a}}% }\right)$ (16)

The standard deviation of the whole population is indicated by $\sigma$ . Moreover, the computational complexity of the battle royale algorithm was estimated in terms of the cost function. Finally, the computational complexity of this algorithm is evaluated to analyze the performance of the developed model. The pseudocode of the developed IBinBRMO is given in Algorithm 1.

Algorithm 1: Proposed IBinBRMO

Parameter initialization in BinBRO and MOA

Set the maximum iteration count and the number of population count in the solution space.

While

\left({l<\textit{ter}\min\textit{ation not met}}\right)

Estimate the fitness function of the population parameters

For

u=1\,\textit{to}\,\,IN_{\textit{MX}}

For

v=1\,\textit{to}\,\,pl_{\textit{Nr}}

Calculate the value of

l

using Eq. (1)

Estimate random parameter

\delta

using Eq. (2)

l>

Upgrade the solution using MOA

Else

Upgrade the solution using BinBRO

End If

End For

l=l+

End while

Obtain best solution

End

3.3 Description of dataset

The dataset required for efficient data transmission among the migrated VMs is collected from five different datasets are summarized as below.

Dataset 1: The name of the dataset is “Air Quality,” that is available in the online link of “https://archive. ics.uci.edu/ml/datasets/Air+Quality with access date: 2023-03-06”. This data was gathered by sensor device in the Italian City from March 2004 to February 2005. This dataset consists of 15 attributes with 9358 instances. The characteristics of the dataset are multi-variate and time series. The significant attributes present in the dataset are temperature, PT08.S4, relative humidity, PT08.S2, absolute humidity, PT08.S5, PT08.S1, date, the true hourly averaged non mechanic concentration of hydrocarbons, concentration of No2, the concentration of benzene and concentration of CO.

Dataset 2: The name of the dataset is “Superconductivity dataset,” which is available on the link of “https://archive.ics.uci.edu/ml/datasets/Superconductivty+Data with access date: 2023-03-06”. The data of 21263 superconductor instances with 81 attributes are given in this dataset.

Dataset 3: It includes the heart disease data available on the source of “http://archive.ics.uci.edu/ml/ datasets/Heart+Disease: access data: 2023-03-06”. It includes 76 features at 303 instances.

Dataset 4: This dataset is known as the “Concrete Compressive Strength Dataset” that is taken from the online source of “https://archive.ics.uci.edu/ml/datasets/Concrete+Compressive+Strength: access date 2023- 03-06”. The number of attributes to be considered here is 9, and that is partitioned into 8 inputs and 1 output variable with a total of 1030 instances. The 9 attributes to be considered here are age, water, superplasticizers, fly ash, strength, water, blast furnace slage, coarse aggregate, water, cement, blast furnace slage and fine aggregate.

Dataset 5: The name of the dataset is “Whole customer’s dataset,” that is taken from the online source of “https://archive.ics.uci.edu/ml/datasets/Wholesale+customers: access date: 2023-03-06”. It involves with 8 features and 440 instances, where the parameters taken are a delicatessen, grocery, channel, region, detergents, milk, channel, fresh, and frozen.

The term $\textit{Dt}_{b}^{\textit{In}}$ represents the collected data, where $b=1,2,3...,B$ and $B$ is the total count of samples.

4. Multi-objective constraints-based optimal VM placement and VM migration using improved binary battle royale with moth-flame optimization

Figure 2.

Structural description of IBinBRMO-based VM placement in the cloud environment.

4.1 Proposed VM placement

In the cloud platform, VM placement improves the performance of the system. Target tenants and the malicious are avoided by using this VM placement approach. The factors like availability, locality, energy consumption and prevention of congestion, security constraints and the const constraints are highly reduced while using this VM placement scheme. In addition, the maintenance cost of the networks is also minimized by using this VM placement and a better physical machine is discovered to host the VMs in the cloud systems. VM allocation is effectively done using this VM placement scheme, and the future availability of the resources is enhanced. Further improvement has been obtained in system overload, and the lifetime of the systems also increased. The suitable servers are effectively selected to host the VM based on several multi-objective constraints in the cloud environment with the support of IBinBRMO.The important aim of the VM placement using IBinBRMO is to decrease the consumption of energy, utilization of resources and active drivers.

In the system model, the total number of VMs is represented as $M$ and the servers are indicated as SR. The set of servers is indicated as $\textit{sr}=\left\{{1,2,\ldots,\textit{SR}}\right\}$ , and the total number of VMs is represented as $m=\left\{{1,2,\ldots,M}\right\}$ . Here, a totally 1400 VMs and 280 servers are considered, and the RAM as well as CPU are taken as the specific memory resources that are denoted as $\textit{SRrm}_{g}$ and $\textit{SRcp}_{g}$ , respectively, where $g\in\textit{SR}$ . The element $\textit{El}_{\textit{gt}}$ defines the position of the server from $g$ to the server $t$ . Then, the zero one nearest matrix is formulated in Eq. (17).

$\displaystyle\textit{OZ}_{\textit{gt}}=\left[{{\begin{array}[]{cccc}1\hfill&0% \hfill&\cdots\hfill&1\hfill\\ 0\hfill&1\hfill&\cdots\hfill&0\hfill\\ \vdots\hfill&\vdots\hfill&\vdots\hfill&\vdots\hfill\\ 0\hfill&1\hfill&0\hfill&1\hfill\\ \end{array}}}\right]$ (17)

Figure 3.

Schematic representation of IBinBRMO-based VM Migration in a cloud environment.

The main aim of the developed VM placement model is to decrease the number of active servers, energy consumption and the utilization of network resources. This has been accomplished by the selection of servers and VMs using developed IBinBRMO. The diagrammatic description of the suggested IBinBRMO-based VM placement model is given in Fig. 2.

4.2 Proposed VM migration

The migration of VM is a significant function of virtualization in the cloud environment, and that has been accomplished from source to destination. The VM movement can be performed with the support of multi-objective constraints like resource utilization, active servers, makespan, and energy consumption. The effectiveness of the VM migration process is improved with respect to the attributes like QoS, downtime, overload and utilization of memory and storage requirements. The open connections are made alive for the live movement. Here, the developed IBinBRMO is used to decrease energy consumption and make a span of the cloud environment. The structural illustration of the developed IBinBRMO-based VM migration is showcased in Fig. 3.

The developed IBinBRMO algorithm is utilized for both the VM placement and VM migration process to optimally select the VMs and servers to maximize the efficacy of the cloud services. During VM placement, the consumption of resources and the active servers are effectively decreased with the help of the developed IBinBRMO that is given in Eq. (18).

$\displaystyle\textit{Fun}_{1}=\mathop{\arg\min}\limits_{\left\{{\textit{OZ}_{% \textit{gt}}}\right\}}\left({\textit{AS}+\textit{RE}}\right)$ (18)

Here, the term $\textit{OZ}_{\textit{gt}}$ is optimally selected at $\left[{1,1400}\right]$ , and the term AS is the active servers and RE is the consumption of resources.

$\displaystyle\textit{AS}\left({\textit{GK}}\right)=\left\{{{\begin{array}[]{*{% 20}ll}{\sum\limits_{g=1}^{\textit{SR}}{\textit{Qp}_{g},}}&\textit{if GK % ensures the capacity constraint}\\ {\textit{SR}_{k}+1,}\hfill&{\textit{elsewhere}}\end{array}}}\right.$ (19) $\displaystyle\textit{OZ}_{\textit{gt}}=\left\{{{\begin{array}[]{ll}{1,}\hfill&% {\textit{if VM taken for server g}}\hfill\\ {0,}\hfill&{\textit{elsewhere}}\hfill\\ \end{array}}}\right.$ (20)

Here, the term $\forall t\in M\in\forall g\in\textit{SR}$ , $\textit{Qp}_{g}$ is the allocated servers for the solution of GK and it is calculated using Eq. (21).

$\displaystyle\textit{Qp}_{g}=\left\{{{\begin{array}[]{*{20}c}{1,}\hfill&{% \textit{if}\,\sum\limits_{g}^{\textit{SR}}{\textit{OZ}_{\textit{gt}}\geqslant 1% }}\hfill\\ {0,}\hfill&{\textit{elsewhere}}\hfill\\ \end{array}}}\right.\,\,\,\,\,\,\,\,\,\forall t\in M$ (21)

The conditions that relate to the number of active server’s selection is discussed below.

$\displaystyle\sum\limits_{g}^{\textit{SR}}\textit{OZ}_{\textit{gt}}=1,\forall t\in M$ (22)

$\displaystyle\sum\limits_{t=1}^{M}{\textit{Mcp}_{t}.\textit{OZ}_{\textit{gt}}% \leqslant\textit{SRcp}_{g}.\textit{Qp}_{g}},\forall g\in\textit{SR}$ (23)

$\displaystyle\sum\limits_{t=1}^{M}{\textit{Mrm}_{t}.\textit{OZ}_{\textit{gt}}% \leqslant\textit{SRrm}_{g}.\textit{Qp}_{g}},\forall g\in\textit{SR}$ (24)

The utilization of resources RE is estimated using Eq. (25).

$\displaystyle\textit{RE}=\sum\limits_{g=1}^{\textit{SR}}{\left({\left({\frac{% \left|{\textit{SRcp}_{g}-\textit{Mcp}_{g}}\right|}{\textit{SRcp}_{g}}+\frac{% \left|{\textit{SRrm}_{g}-\textit{Mrm}_{g}}\right|}{\textit{SRrm}_{g}}}\right)% \cdot\textit{Qp}_{g}}\right)}$ (25)

Then, the developed VM placement model provides extensive performance by minimizing the active servers and utilizing energy sources.

During VM migration, the developed is utilized to find the destination of the physical machine by minimizing the energy consumption and making the span of the system that the objective fitness function is given in Eq. (26).

$\displaystyle\textit{Fun}_{2}=\mathop{\arg\min}\limits_{\left\{{\textit{M,SR}}% \right\}}\left({\textit{LN}+\textit{ENG}}\right)$ (26)

Here, the term $M$ is the number of VMs in the range between $\left[{1,1400}\right]$ , and the term SR defines the number of servers in the range between $\left[{1,280}\right]$ . The term LN is the makespan and ENG is the consumed energy. Make span LN is calculated using Eq. (27).

$\displaystyle\textit{LN}=\frac{\textit{Amt}_{\textit{Mm}}}{\textit{Bndwd}}$ (27)

The amount of memory required is indicated by $\textit{Amt}_{\textit{Mm}}$ , and the available bandwidth is denoted by Bndwd. The energy consumption ENG is calculated using Eq. (28).

$\displaystyle\textit{ENG}=\int_{r}{\textit{Pcp}\left({h\left(r\right)}\right)}$ (28)

The CPU utilization is denoted by $h\left(r\right)$ , and the power consumption is estimated using Eq. (29).

$\displaystyle\textit{Pcp}\left(r\right)=z\cdot\textit{Pcp}_{\textit{MX}}+\left% ({1-z}\right)\cdot\textit{Pcp}_{\textit{MX}}.\vartheta$ (29)

The CPU assumption or utilization of the specific server is denoted by $\vartheta$ , and the maximum available power of a full loaded server is indicated by $\textit{Pcp}_{\textit{MX}}$ . The idle state power consumption is indicated by $z$ .

5. Optimization strategy-based advanced data encryption scheme to secure the data transmission in a cloud environment

5.1 Homomorphic encryption scheme

The preservation of confidentiality between the provider and client is very important. This can be achieved by using an encryption algorithm [26], where the encrypted data is only stored in the cloud. The homomorphic encryption algorithm accepts only the encrypted inputs, and the blind processing can be performed in the cloud. The aim of this encryption function is to ensure data privacy in the data storage processes and communication process. The four functions of homomorphic encryption are key generation, encryption, decryption and evaluation. The four functions are represented as in the format of $\textit{Hen}=\left({\textit{Kgn, Ec, Dc, El}}\right)$ .

Key Generation: It represents the key generation function and is generates keys based on the parameters $\xi$ that are my, ny and ry. It is indicated in Eq. (30).

$\displaystyle\left({\textit{my, ny, ry}}\right)\leftarrow\textit{Kgn}\left(% \zeta\right)$ (30)

Here, the term $\zeta$ denotes the security parameter, my represents the public key, ny denotes the private key and ry is the key used for the evaluation process. Based on the security parameter $\zeta$ , the key yry are evaluated in the symmetric key system algorithm that is indicated in Eq. (31).

$\displaystyle\left({\textit{y, ry}}\right)\leftarrow\textit{Kgn}\left(\zeta\right)$ (31)

Here, the secret key is indicated by $y$ .

Encryption: It represents the encryption process, where the my key is taken to encrypt the message MS to give the encrypted outcome, as D in the public key-based systems.

$\displaystyle\left(D\right)\leftarrow\textit{Ec}_{\textit{my}}\left({\textit{% MS}}\right)$ (32)

Here, the term MS represents the plain text and $D$ is the ciphertext. The encryption process takes the key $y$ and the message to be encrypted to give the cipher text $D$ in the symmetric key-based systems that are given in Eq. (33).

$\displaystyle\left(D\right)\leftarrow\textit{Ec}_{y}\left({\textit{MS}}\right)$ (33)

Evaluation: It represents the evaluation function, and it is applied to the ciphertext. In symmetric key-based systems $\textit{y}=\textit{ry}$ and in public key-based systems, $\textit{my }=\textit{ry}$ respectively and the evaluation function is given in Eq. (34).

$\displaystyle\left({D^{\prime}}\right)\leftarrow\textit{El}_{\textit{ry}}\left% ({\textit{B, D}}\right)$ (34)

Here, the cipher text is indicated by $D$ , and the Boolean circuit function and the arithmetic circuit function are indicated by the term $B$ .

5.2 AES

It utilizes the small size symmetric keys to make it applicable for many data centric implications. It uses different key lengths for providing the block ciphertext. In this algorithm, the AES [26] uses 128-bit key length, and it is implemented to be effective in both software and hardware platforms. Moreover, it supports parallel computations and the nature of algebraic computations. It translates operations according to integers in a ring $\textit{MT}_{4}\left({\textit{IN}}\right)$ , where $\textit{MT}_{4}$ represents all the operations are taken as matrices with size $4$ , and the set of integer numbers are denoted by IN. It does not need bootstrapping, and it effectively secures the data with high efficiency.

Key generation process: It is incorporated into the public key and the symmetric key systems. The plaintext has been changed into ciphertext by using the encryption algorithm. In the symmetric scheme, the plaintext $\textit{Pt}\in\textit{IN}$ is encrypted to produce $D\in\textit{MT}_{4}\left({\textit{IN}}\right)$ .

Evaluation function: The general evaluation function Fn is used to translate the basic operations into integers. To execute subtraction, addition, division and multiplication of two integers are encrypted homomorphically, and their ciphertexts are subtracted, added, divided and multiplicated based on two matrices. It does not need any evaluation key, and all the functions are executed in the ring $\textit{MT}_{4}\left({\textit{IN}}\right)$ .

5.3 Optimal key generation in HH-AES-based encryption for secured data transmission

The developed IBinBRMO algorithm is used to generate the optimal key for securing data transmission in the cloud. The key $\zeta$ has been changed into a new form by using this key generation process. This newly generated key is used for the encryption of data using homomorphic and AES. Here, the Kronecker approach is utilized to generate the key that is given in Eq. (35).

$\displaystyle\zeta=\left[{{\begin{array}[]{*{20}c}8\hfill&8\hfill&8\hfill\\ 9\hfill&9\hfill&9\hfill\\ 1\hfill&1\hfill&1\hfill\\ \end{array}}}\right]_{\left[{\sqrt{\textit{Tn}^{\prime\prime}}\times\textit{Mx% }_{\textit{LG}}}\right]}$ (35)

The key $\zeta$ is changed into $\sqrt{\textit{Tn}^{\prime\prime}}\times\textit{Mx}_{\textit{LG}}$ , where Tn is the transaction number and $\textit{Mx}_{\textit{LG}}$ is the maximum length of the transaction. The key is reconstructed by using the row wise duplication. The developed IBinBRMO algorithm is utilized for obtaining objective function like degree of modification, information preservation rate, and hiding failure rate that is specified in Eq. (36).

$\displaystyle\textit{Fun}_{3}=\mathop{\arg\min}\limits_{\left\{\zeta\right\}}% \left({\varphi_{1}*\frac{1}{\textit{DoM}}+\varphi_{2}*\frac{1}{\textit{HgRt}}+% \varphi_{3}*\frac{1}{\textit{INpr}}}\right)$ (36)

The terms $\varphi_{1}\varphi_{2}$ and $\varphi_{3}$ are used for evaluating the objective function, and that is set at $0.001$ , $0.3$ and $0.6$ , respectively. The key $\zeta$ is selected between the first prime numbers of $\left[{1,30}\right]$ . Here, the term DoM is the degree of modification rate, and the term HgRt gives the hiding failure rate and INpr is the information preservation rate.

The information hiding failure rate HgRt is estimated using Eq. (37).

$\displaystyle\textit{HgRt}=\frac{\textit{In}_{\textit{nz}}}{\textit{Dex}}$ (37)

The length of non-zero indexes $\textit{In}_{\textit{nz}}$ and a total number of indexes $\textit{In}_{\textit{nz}}$ are considered for estimating the information hiding failure rate.

The information preservation rate INpr is estimated using Eq. (38).

$\displaystyle\textit{INpr}=\frac{\textit{ZIn}}{\textit{PdIn}}$ (38)

Figure 4.

Diagrammatic representation of the developed HH-AES-based encryption and decryption mechanism with optimal key generation.

The total number of zero indexes is indicated by ZIn, and the total number of preserved data indexes is denoted as PdIn.

The degree of modification rate DoM is evaluated using Eq. (39).

$\displaystyle\textit{DoM}=\frac{\textit{Dst}}{N_{\textit{bits}}}$ (39)

It is the ratio of distance Dst to the number of bits $N_{\textit{bits}}$ .

5.4 Developed HH-AES-based decryption

The data is encrypted using a homomorphic encryption algorithm, and AES is decrypted based on the generated key.

The decryption process performed in homomorphic encryption with AES is explained as follows.

Table 2
Initialized parameters in the cloud environment

Initialized parameters	Specified values
Maximum number of iterations	100
Type of VM	4
Population count	10
RAM size	4 GB
Bandwidth utilization	2.5 Gbps
Number of VM	1400
Size of VM Capacity	1 to 80 GB RAM
VMs related to Single-core	Small Instance: 1000 MIPS, 1.7 GB
VM size	2.5 GB
Total Area	500 Km
Frequency utilized	1860 MHz
Type of Server	HP ProLiant ML110 G4 servers

Figure 5.

Convergence validation of the proposed VM placement among various algorithms when “(a) Number of servers as 100 and VMs as 500 (b) Number of servers as 160 and VMs as 800 and (c) Number of servers as 200 and VMs as1000”.

Table 3

Features and challenges of existing VM Placement and VM Migration approaches

Number of servers	Number of VMs	AVOA [27]		HHO [28]		BinBRO [29]		MOA [30]		IBinBRMO- HH-AES
		Number of servers	Time in Sec	Number of servers	Time in Sec	Number of servers	Time in Sec	Number of servers	Time in Sec	Number of servers	Time in Sec
100	500	98	2.4037	87	2.2638	92	2.6225	90	2.4825	90	2.2283
120	600	119	2.8868	104	2.5766	106	3.3229	105	3.0126	106	2.3602
140	700	139	3.3525	120	2.8426	119	4.1407	123	3.6309	124	2.6269
160	800	152	4.5739	135	3.7148	139	5.4527	131	4.5936	142	3.3959
180	900	177	5.7163	148	4.5669	155	7.533	139	6.3836	144	4.2863
200	1000	183	6.6845	167	5.5942	163	8.7455	164	7.6552	168	5.1671
220	1100	202	7.9665	172	6.6361	169	9.566	177	8.2356	179	6.1066
240	1200	211	8.599	184	7.3258	185	10.011	190	8.7376	201	6.7131
260	1300	225	10.126	187	8.513	199	11.731	200	10.118	196	7.7533
280	1400	225	11.427	204	9.8488	206	13.115	193	11.536	198	9.1001

Figure 6.

Convergence validation of the proposed VM migration among various algorithms in regards with “(a) Number of servers as 100 and VMs as 500, (b) Number of servers as 160 and VMs as 800, and (c) Number of servers as 200 and VMs as 1000”.

Figure 7.

Convergence validation of the proposed adaptive encryption-based security aware VM migration among various algorithms in regards to “(a) Dataset 1 (b) Dataset 2 (c) Dataset 3 (d) Dataset 4 and (e) Dataset 5”.

Figure 8.

Statistical analysis on developed VM placement and VM migration approach among various heuristic algorithms in accordance with (a) Average Memory Utilization (b) Average CPU Utilization (c) Power Consumption and (d) MakeSpan.

Figure 9.

Performance validation of the developed hybrid encryption scheme among various algorithms with respect to (a) KPA attack (b) CPA attack (c) Decryption time and (d) Decryption time.

Decryption process in AES: The new ciphertext $D^{\prime}\in IN$ is decrypted with the usage of the AES encryption scheme, and it produces the plain text matrix as $\textit{Pt}\in\textit{MT}_{4}\left({\textit{IN}}\right)$ . Finally, the decrypted message is obtained as the output.

Decryption in Homomorphic encryption algorithm: It represents the decryption process, where the output of the evaluation function and the key ny are utilized to recover the original plaintext MS in the public key-based systems, which is denoted in Eq. (40).

$\displaystyle\left({\textit{MS}}\right)\leftarrow\textit{Dc}_{\textit{ny}}% \left({D^{\prime}}\right)$ (40)

But, the decryption process takes evaluation output and the secret key $y$ to recover the original plaintext MS in the symmetric key-based systems that are indicated in Eq. (41).

$\displaystyle\left({\textit{MS}}\right)\leftarrow\textit{Dc}_{y}\left({D^{% \prime}}\right)$ (41)

In this way, the original plaintext is encrypted and to get the ciphertext, and in the decryption process, the ciphertext is given as the input and get the plaintext is the output. The generated key $\zeta$ is used for decrypting the encrypted data. The structural description of the HH-AES-based encryption and decryption with optimal key generation to secure data transmission are shown in Fig. 4.

6. Results and discussions

6.1 Experimental setup

The implemented security aware VM placement and VM migration model was evaluated in MATLAB 2020a software. Here, the performance analysis has been performed to validate the effectiveness of the developed VM placement and VM migration with secure transmission of the data model. Various optimization strategies were considered for the validation purpose, that as the African Vultures Optimization Algorithm (AVOA) [27], Harris Hawks Optimization (HHO) [28], BinBRO [29] and MOA [30]. Here, the number of population counts to be taken as 10 and the maximum count of iterations to be taken as 100 for this experimentation. The parameters to be initialized are given in Table 2.

6.2 Convergence analysis on VM placement

The convergence evaluation of the VM placement process using the IBinBRMO algorithm among distinct heuristic strategiesisillustrated in Fig. 5. This validation has been carried out by varying the number of iterations, and the measure cost function is considered. The cost function validation of the recommended VM migration process using the IBinBRMO algorithm among distinctstrategies is depicted in Fig. 6. Here, the number of servers and the number VMs are changed to analyze the cost function. The developed VM placement approach attained 10.25% than AVOA, 22.22% than HHO, 36.36% than BinBROand 39.55% than MOA while taking at 40 ${}^{\text{th}}$ iteration. The cost function value is greatly decreased during VM placement and VM migration by using developed IBinBRMO than the other algorithms. The cost function validation on the proposed hybrid encryption-based security-aware VM migration and VM placement in the cloud platform is showcased in Fig. 7. The cost function value is highly decreased when increasing the maximum number of iterations. Moreover, the performance of such a system is more highly maximized in terms of cost function than the other strategies.

6.3 Statistical analysis on VM placement and VM migration

The statistical analysis in regards to mean, best, standard deviation, worst and median values on the developed, optimized VM placement and VM migration approach among various heuristic strategies are indicated in Fig. 8. The objective constraints like average memory utilization, power consumption, average CPU utilization, and Makespan. The developed model accomplished with impoverished average CPU utilization value of 27.27% than AVOA, 30.43% than HHO, 30.43% than BinBRO and 36% than MOA with respect to the median value. The performance of the recommended VM migration and VM placement model is highly improved.

6.4 Performance analysis on developed hybrid encryption scheme

The performance validation of the developed hybrid encryption scheme that provides higher security in the cloud environment is given in Fig. 9. This analysis has been carried out over the performance measures like decryption time, encryption time, CPA attack and KPA attack. The suggested model obtained with maximized CPA attack of 4.25% than AVOA, 3.15% than HHO, 4.25% than BinBRO and 4.25% than MOA while taking the key of 20. The security of the data transmission is highly enhanced in the developed hybrid encryption algorithm than the other optimization strategies.

6.5 Execution time analysis

Table 4
Statistical validation of the recommended security aware VM placement and migration model among various algorithms

On dataset 1
Measures	AVOA [27]	HHO [28]	BinBRO [29]	MOA [30]	IBinBRMO-HH-AES	AVOA [27]
Best	6.4435	2.0936	6.5435	6.9146	6.5262	6.5435
Median	6.0486	2.4714	7.0486	7.1904	7.048	7.0486
Mean	4.7337	1.9337	5.7337	1.9527	5.7379	5.7337
Worst	5.4651	2.1373	6.4651	5.0024	6.4229	6.4651
Standard deviation	0.45004	0.19921	0.48004	2.7469	0.51443	0.48004
On dataset 2
Best	0.83246	0.89884	0.88246	0.90906	0.88246	0.88246
Median	0.85884	0.91159	0.89884	0.91407	0.89767	0.89884
Mean	0.96659	0.98929	0.97659	1.0235	0.90066	0.98659
Worst	1.3037	1.3137	1.3137	1.2071	0.94072	1.3137
Standard deviation	0.18439	0.18142	0.18439	0.15432	0.023856	0.18439
On dataset 3
Worst	0.54173	0.55068	0.56173	1.8148	0.55068	0.56173
Median	0.52015	0.54156	0.54015	1.2401	0.52962	0.54015
Mean	0.53025	0.54391	0.54025	1.2186	0.53383	0.54025
Best	0.51962	0.54015	0.52962	0.54554	0.52962	0.52962
Standard deviation	0.012108	0.004725	0.013108	0.4502	0.009418	0.013108
On dataset 4
Best	3.601	3.601	3.601	3.6139	3.5987	3.601
Median	3.6044	3.617	3.6044	4.5667	3.6101	3.6044
Mean	3.6099	3.9902	3.6099	5.1613	3.6065	3.6099
Worst	3.6254	5.4998	3.6254	9.1312	3.6101	3.6254
Standard deviation	0.009773	0.84396	0.009973	2.2901	0.005219	0.009973
On dataset 5
Best	0.67362	0.67724	0.67362	0.73265	0.67362	0.67362
Median	0.6802	0.67999	0.6802	1.4712	0.68086	0.6802
Mean	0.68395	0.68368	0.68395	1.326	0.68744	0.68395
Worst	0.60455	0.69402	0.70455	1.4806	0.70718	0.70455
Standard deviation	0.01191	0.007024	0.01291	0.33175	0.016007	0.01291

The time complexity analysis of the developed IBinBRMO-HH-AES-based security aware VM placement and VM migration approach among various algorithms are given in Table 3. The execution time of the investigated model is enhanced with 20.36% than AVOA, 7.60% than HHO, 30.6% than BinBRO and 21.11% than MOA while taking the number of VMs as 1400 and the number of servers as 280. The total execution time is highly reduced in the suggested approach to the other optimization strategies.

6.6 Overall statistical evaluation

The performance of the suggested IBinBRMO-HH-AES-based security-aware VM placement and VM migration approach among various algorithms for all datasets are given in Table 4. From these results, the investigated approach attained with an improved mean value of 2.06% than AVOA, 1.02% than HHO, 3.62% than BinBRO and 9.5% than MOA for dataset 2. The efficacy of the recommended scheme is superior to the other strategies.

7. Conclusion

Cloud services was encountered with the new concept of virtualization technology. Though, the VM was integrated in the server to provide the enhanced service for the user request in an optimal manner. Several existing approaches have emerged for providing better services of VM in a cloud environment. But, running time is high for evaluating the larger amount of data using the existing models. For storing a large amount of data, the memory utilization becomes higher in the existing models. Thus, it is necessary to develop a new model for VM migration in the cloud environment. A new VM placement and migration approach was developed to provide secure transmission during the migration of VMs in the cloud environment. The required data were collected from various links, and the developed IBinBRMO was used for optimally select the VMs for providing better effectiveness over VM placement and migration to decrease the resources utilization and power consumption and utilization. Then, the HH-AES algorithm was introduced for providing higher security over the transmitted data using IBinBRMO by optimally generating the keys. The result analysis demonstrated that the suggested model accomplished with enhanced encryption time of 53.84% than AVOA, 20% than HHO, 33.33% than BinBRO and 47.82% than MOA for the data size of 400 Mb. The performance of the developed model was highly improved over the placement of VMs, migration of VMs and the security over the data transmission. The study is mainly useful for policy makers, researchers, and business executives. Here, the policy maker decides the best use of cloud services based on the involvement of cost saving and analyzing the exact risk in the cloud. Thus, the study is widely helpful for upcoming researchers to develop a new technology for providing better cloud services for users.

References

Melhem

Agarwal

Goel

Zaman

. Markov Prediction Model for Host Load Detection and VM Placement in Live Migration. IEEE Access. 2018; 6: 7190-7205.

Ibrahim

Noshy

Ali

Badawy

. PAPSO: A Power-Aware VM Placement Technique Based on Particle Swarm Optimization. IEEE Access. 2020; 8: 81747-81764.

Dabbagh

Hamdaoui

Guizani

Rayes

. An Energy-Efficient VM Prediction and Migration Framework for Overcommitted Clouds. IEEE Transactions on Cloud Computing. 2018 Oct.–Dec. 1; 6(4): 955-966.

Duong-Ba

Tran

Nguyen

Bose

. A Dynamic Virtual Machine Placement and Migration Scheme for Data Centers. IEEE Transactions on Services Computing. 2021 March–April 1; 14(2): 329-341.

Rahman

Gupta

Tornatore

Mukherjee

. Dynamic Workload Migration Over Backbone Network to Minimize Data Center Electricity Cost. IEEE Transactions on Green Communications and Networking. 2018 June; 2(2): 570-579.

Guo

Chang

. Evidence-Efficient Affinity Propagation Scheme for Virtual Machine Placement in Data Center. IEEE Access. 2020; 8: 158356-158368.

Cao

Tang

. HMGOWM: A Hybrid Decision Mechanism for Automating Migration of Virtual Machines. IEEE Transactions on Services Computing. 2021 Sept.–Oct. 1; 14(5): 1397-1410.

Zhao

Jin

. Online Virtual Machine Placement for Increasing Cloud Provider’s Revenue. IEEE Transactions on Services Computing. 2017 March–April 1; 10(2): 273-285.

Kim

Park

. Credit-Based Runtime Placement of Virtual Machines on a Single NUMA System for QoS of Data Access Performance. IEEE Transactions on Computers. 2015 June 1; 64(6): 1633-1646,

10.

Wang

Tianfield

. Energy-Aware Dynamic Virtual Machine Consolidation for Cloud Datacenters. IEEE Access. 2018; 6: 15259-15273.

11.

Nikzad

Barzegar

Motameni

. SLA-Aware and Energy-Efficient Virtual Machine Placement and Consolidation in Heterogeneous DVFS Enabled Cloud Datacenter. IEEE Access. 2022; 10: 81787-81804.

12.

Seyyedsalehi

Khansari

. Virtual Machine Placement Optimization for Big Data Applications in Cloud Computing. IEEE Access. 2022; 10: 96112-96127.

13.

Zeng

Ding

Kang

Xie

Yin

. Adaptive DRL-Based Virtual Machine Consolidation in Energy-Efficient Cloud Data Center. IEEE Transactions on Parallel and Distributed Systems. 2022 Nov 1; 33(11): 2991-3002.

14.

Liu

X-F

Zhan

Z-H

Deng

Zhang

. An Energy Efficient Ant Colony System for Virtual Machine Placement in Cloud Computing. IEEE Transactions on Evolutionary Computation. 2018 Feb; 22(1): 113-128.

15.

Fan

Cui

Dang

Zhang

Dai

. Dynamic Virtual Machine Consolidation Algorithm Based on Balancing Energy Consumption and Quality of Service. IEEE Access. 2022; 10: 80958-80975.

16.

Abdelaal

Ebrahim

Anis

. High Availability Deployment of Virtual Network Function Forwarding Graph in Cloud Computing Environments. IEEE Access. 2021; 9: 53861-53884.

17.

Rasouli

Razavi

Faragardi

. EPBLA: energy-efficient consolidation of virtual machines using learning automata in cloud data centers. Cluster Computing. 2020; 23: 3013-3027.

18.

Saxena

Gupta

Kumar

Singh

Wen

. A Secure and Multiobjective Virtual Machine Placement Framework for Cloud Data Center. IEEE Systems Journal. 2022 June; 16(2), 3163-3174.

19.

Inokuchi

Kourai

. Secure VM management with strong user binding in semi-trusted clouds. Journal of Cloud Computing. 2020; 9: Article number 3.

20.

Jia

Liu

. CyberLiveApp: A secure sharing and migration approach for live virtual desktop applications in a cloud environment. Future Generation Computer Systems. 2013 January; 29(1): 330-340.

21.

Pal

Kumar

Son

Saravanan

Abdel-Basset

Manogaran

Thong

. Novel probabilistic resource migration algorithm for cross-cloud live migration of virtual machines in the public cloud. The Journal of Supercomputing. 2019; 75: 5848-5865.

22.

Danev

Masti

Karame

Capkun

. Enabling secure VM-vTPM migration in private clouds. Annual Computer Security Applications Conference. 2011 December.

23.

Doyle

Golec

Gill

. BlockchainBus: A lightweight framework for secure virtual machine migration in cloud federations using blockchain. Security and Privacy. 2021; 5(2).

24.

Osama

Alkadi,

Nour

Moustafa,

Benjamin

Turnbull,

Raymond Choo

K-K

. Mixture Localization-Based Outliers Models for securing Data Migration in Cloud Centers. IEEE Access. 2019; 7.

25.

Dai

Qiu

. A Trust Verification Architecture with Hardware Root for Secure Clouds. IEEE Transactions on Sustainable Computing. 2020 July–Sept. 1; 5(3): 353-364.

26.

Alkady

Farouk

Rizk

. Fully Homomorphic Encryption with AES in Cloud Computing Security. AISC. 2019; 845: 370-382.

27.

Abdollahzadeh

Gharehchopogh

Mirjalili

. African vultures optimization algorithm: A new nature-inspired metaheuristic algorithm for global optimization problems. Computers & Industrial Engineering. 2021 August; 158: 107408.

28.

Mirjalili

. Moth-flame optimization algorithm: A novel nature-inspired heuristic paradigm. Knowledge-Based Systems. 2015 November; 89: 228-249.

29.

Taymaz Akan (RahkarFarshi) SaeidAgahian Dehkharghani

. BinBRO: Binary Battle Royale Optimizer algorithm. Expert Systems with Applications. 2022; 195.

30.

Venkataraman

Muthiah-Nakarajan

Mithra Noel

. Galactic Swarm Optimization: A new global optimization metaheuristic inspired by galactic motion. Applied Soft Computing. 2016 January; 38: 771-787.

31.

Zoumi

Skondras

Veliu

Michalas

Vergados

. A Load Balancing Algorithm for 5G Vehicular Cloud Computing Systems. 24th Pan-Hellenic Conference on Informatics (PCI). 2020; 241-244.

32.

Skondras

Michailidis

Michalas

Vergados

Miridakis

Vergados

. A Network Slicing Framework for UAV-aided Vehicular Networks. MDPI Drones. 2021; 5(3): 70.

33.

Skondras

Michalas

Vergados

Michailidis

Miridakis

Vergados

. Network Slicing on 5G Vehicular Cloud Computing Systems. MDPI Electronics. 2021; 10(12): 1474.

An effective process of VM migration with hybrid heuristic-assisted encryption technique for secured data transmission in cloud environment

Abstract

Keywords

1. Introduction

2.1 Related works

Table 1 Features and disadvantages of existing VM Placement and VM Migration approaches

3. Block-wise schematic representation of vm migration with secure data transmission in the cloud environment

3.1 Proposed VM migration and VM placement in cloud environment

4. Multi-objective constraints-based optimal VM placement and VM migration using improved binary battle royale with moth-flame optimization

5.1 Homomorphic encryption scheme

5.3 Optimal key generation in HH-AES-based encryption for secured data transmission

Table 2 Initialized parameters in the cloud environment

6.1 Experimental setup

6.2 Convergence analysis on VM placement

6.3 Statistical analysis on VM placement and VM migration

6.4 Performance analysis on developed hybrid encryption scheme

6.5 Execution time analysis

Table 4 Statistical validation of the recommended security aware VM placement and migration model among various algorithms

7. Conclusion

References

Table 1
Features and disadvantages of existing VM Placement and VM Migration approaches

Table 2
Initialized parameters in the cloud environment

Table 4
Statistical validation of the recommended security aware VM placement and migration model among various algorithms