
==== Front
Sci Rep
Sci Rep
Scientific Reports
2045-2322
Nature Publishing Group UK London

39232132
71682
10.1038/s41598-024-71682-y
Article
Fog-assisted de-duplicated data exchange in distributed edge computing networks
Said Ghawar 17
Ghani Anwar anwar.ghani@iiu.edu.pk

12
Ullah Ata 3
Alzahrani Abdulrahman 4
Azeem Muhammad 13
Ahmad Rashid 25
Kim Do-Hyeun kimdh@jejunu.ac.kr

26
1 https://ror.org/047w75g40 grid.411727.6 0000 0001 2201 6036 Department of Computer Science, International Islamic University, Islamabad, 44000 Pakistan
2 https://ror.org/05hnb4n85 grid.411277.6 0000 0001 0725 5207 Big Data Research Center, Jeju National University, Jeju-si, Jeju-do 63243 South Korea
3 https://ror.org/008dh2426 grid.444798.2 0000 0004 0607 5732 Department of Computer Science, National University of Modern Languages, Islamabad, 44000 Pakistan
4 https://ror.org/015ya8798 grid.460099.2 0000 0004 4912 2893 Department of Information Systems and Technology, University of Jeddah, 21959 Jeddah, Saudi Arabia
5 https://ror.org/02ftvf862 grid.444763.6 0000 0004 0427 5968 Faculty of Computing and Information Technology, Sohar University, 311 Sohar, Sultanate of Oman
6 https://ror.org/05hnb4n85 grid.411277.6 0000 0001 0725 5207 Department of Computer Engineering, Jeju National University, Jeju-si, Jeju-do 63243 South Korea
7 https://ror.org/0254sa076 grid.449131.a 0000 0004 6046 4456 Department of Computing and Technology, Iqra University, H-9 Campus, Islamabad, 44000 Pakistan
4 9 2024
4 9 2024
2024
14 2059513 9 2023
29 8 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
The Internet of Things (IoT) generates substantial data through sensors for diverse applications, such as healthcare services. This article addresses the challenge of efficiently utilizing resources in resource-scarce IoT-enabled sensors to enhance data collection, transmission, and storage. Redundant data transmission from sensors covering overlapping areas incurs additional communication and storage costs. Existing schemes, namely Asymmetric Extremum (AE) and Rapid Asymmetric Maximum (RAM), employ fixed and variable-sized windows during chunking. However, these schemes face issues while selecting the index value to decide the variable window size, which may remain zero or very low, resulting in poor deduplication. This article resolves this issue in the proposed Controlled Cut-point Identification Algorithm (CCIA), designed to restrict the variable-sized window to a certain threshold. The index value for deciding the threshold will always be larger than the half size of the fixed window. It helps to find more duplicates, but the upper limit offset is also applied to avoid the unnecessarily large-sized window, which may cause extensive computation costs. The extensive simulations are performed by deploying Windows Communication Foundation services in the Azure cloud. The results demonstrate the superiority of CCIA in various metrics, including chunk number, average chunk size, minimum and maximum chunk number, variable chunking size, and probability of failure for cut point identification. In comparison to its competitors, RAM and AE, CCIA exhibits better performance across key parameters. Specifically, CCIA outperforms in total number of chunks (6.81%, 14.17%), average number of chunks (4.39%, 18.45%), and minimum chunk size (153%, 190%). These results highlight the effectiveness of CCIA in optimizing data transmission and storage within IoT systems, showcasing its potential for improved resource utilization and reduced operational costs.

Subject terms

Computer science
Energy science and technology
issue-copyright-statement© Springer Nature Limited 2024
==== Body
pmcIntroduction

Internet of Things (IoT) is a network of smart devices having applications in healthcare1, education2, sports3, transportation4, and smart energy managment5. In IoT, smart devices can communicate with others to share sensed data from their environment, receive data, and distribute information as needed6. IoT will involve billions of communication devices extending the cyber world’s boundaries using real-world objects and digital components. The monitoring sensors node attached to the patient’s body can monitor health parameters like temperature, blood pressure, and heartbeat to forward data towards the fog server7. The fog node extracts the patient information for analysis and processing. After that, the fog node sends the processed information to the cloud node for storage and provides medical professionals access.8. IoT-enabled WSNs can sense, process, and communicate information through communication media. IoT is the driving force behind the IoT-assisted WSN. It allows any bright object to speak with each other, addressing this novel concept of IoT as anything, everywhere, and at any time. The Internet-connected sensor nodes, RFID tags, and near-field communication (NFC) gadgets can detect status and use real-time data. They also have access to past data and have built algorithms, which could include “triggering devices”. It can result in powerful smart settings, such as buildings, healthcare, monitoring systems, and so on9,10.

IoT sensor networks can resolve the physical world’s information-gathering dilemma for humans and link the objective physical world to the human logical society11. The IoT-enabled system often comprises many monitoring sensor nodes combining sensing, communication, processing, and storage capabilities. Multi-hop wireless communication is used to establish it better. The main aim of the monitoring sensor networks is to keep track of the target data inside the coverage region and transmit it to the “Sink node” or terminal users. The monitoring sensor node’s communication, calculation, and power supply are limited due to its size and cost constraints12,13.

For many applications, one of the essential components is thought to be accurate data measurement. Acquiring a comprehensive understanding of measurement data is increasingly important to guarantee a positive user experience. In the real-world scenario, the healthcare temperature, heart-beat, BP, and ECG values may remain within the prescribed range. Still, the sensors keep on sharing the redundant data which increases storage and communication overhead. It can be mitigated through de-duplication as explored in this work14.

The data generated by a single sensor may be insignificant in big data IoT-enabled WSN, such as monitoring sensors that continuously collect and share medical images from different patient wards or outside regions. The data of many monitoring sensor nodes cannot be ignored as they collect data of critical patients with severe symptoms even from the emergency wards15. Energy-aware cloud workflow scheduling has become a hot topic due to cloud data centers’ massive energy demand growth to handle diverse workflow applications16. A method of data minimization is data de-duplication, which removes data redundancy, resulting in it being widely used to conserve storage space and network bandwidth by just keeping one duplicate of the data. De-duplication is a data compression technique used in cloud storage systems to delete redundant data and lower network transmission rates and storage space17,18. The data De-duplication techniques face four issues in IoT-enabled WSNs. The first one is that the monitoring sensors or wearable carry minimal resources, including less storage, which can be over-occupied due to duplicated data storage at most of the sensors in the deployment area. The second one is that the IoT-enabled WSN also suffers excessive energy consumption due to the extra messaging for redundant data. By applying De-duplication, the number of messages is reduced, which results in less energy consumption for transmission. Thirdly, effective bandwidth utilization is critical for the monitoring sensors, and more unique messages can be shared within a specified bandwidth limit. Fourthly, the network traffic can also be handled with limited resources19. The present intelligent transportation system’s (ITS) low latency needs are hard to guarantee in the traditional cloud-based Internet of Vehicles (IoV) architecture. Fog computing is a valuable addition to cloud computing that can help relieve bandwidth and processing resource bottlenecks while enhancing quality of service (QoS) in the IoV20.

Fog computing is a revolutionary paradigm for addressing such difficulties, and it plays a crucial role in large IoT systems for managing real-time data. Fog computing is essential for reducing latency in time-sensitive applications and is required to develop more expert awareness and intelligent decisions21,22. De-duplication at the file level recognizes files with different names but identical content. These file types are removed and replaced with numerical or linked values. The block-level de-duplication divides the data stream into fixed-length or variable-length to identify the same data blocks. The blocks are then matched with their stored data blocks. Byte-by-byte comparison is used in byte-level de-duplication to compare the data stream to the stored data stream23. Big data has a large volume, velocity, and variety and is challenging to collect, process, and store using traditional data collection methods.

Furthermore, the sensors may be densely and extensively spread throughout the monitoring area, resulting in high spatial-temporal correlation and plenty of data redundancy among the detected values. As a result, fusing those felt big data and eliminating redundancy among the vast volume and variety of sensed data is vital24,25. Fig. 1 depicts a general FoG-based system with a three-layer healthcare architecture. The Devices/Users layer, the FoG layer, and the Cloud layer comprise the three-layer architecture. All data is collected from multiple wearable sensor devices in the devices/users layer. When it has been aggregated, it communicates with the FoG layer through a communication medium such as WSN. The FoG layer has many FoG nodes that deliver various services in response to diverse IoT device requests. The Cloud layer provides solutions for many shared resources (computational resources, storage, services, and applications) that support IoT-based systems with consistency and scalability. In healthcare, monitoring is carried out by attaching/implanting resource-limited medical sensor nodes to the patient’s body to transmit data to a sink, a node with abundant resources26.Figure 1 General FoG-Based Healthcare Architecture.

In IoT environments, redundant data storage in the Fog (FoG) or cloud consumes significant storage resources. Thus, it is critical to eliminate redundant values before storing or exchanging information. Data de-duplication focuses on processing data chunks, and multiple de-duplication schemes employing chunking algorithms are explored to optimize this process effectively. The motivation behind the proposed technique stems from the extensive array of storage devices and the substantial surge in data from various sources today, including commercial transactions, social networks, and numerous other domains. Searching tiny data in such a plethora of data is a pivotal topic in computer science. The objective is to eradicate space-consuming files, reduce storage costs, and establish a groundwork for additional space in the system through data de-duplication methods. The data de-duplication can also enhance communication performance by reducing the number of data bits/bytes exchanged among network nodes, leading to low energy consumption and prolonging the network lifetime.

This paper presents a solution to address the weaknesses of the existing schemes. It uses a threshold limit to specify the size of variable_ sized_window that should be larger than the fixed_size_window. The article’s primary contributions are listed below.To introduce a novel CCIA designed to significantly reduce data size in healthcare parameters received from the Aggregator Node within a specified region. The CCIA is an efficient tool for eliminating redundant data, contributing to enhanced data transmission efficiency and overall network optimization.

To use a control cut-point identification algorithm for the data de-duplicated scheme (DDS) in “IoT-enabled WSN”, which dominates the size of the variable size window. The better identification of cut-points can eliminate the transmission of redundant data from various IoT devices by the aggregator node, thereby contributing to the prolonged network life.

To implement the proposed scheme focusing on minimum, maximum, and average chunk sizes and conducting a comprehensive analysis of the scheme’s performance in various scenarios to assess its effectiveness in reducing redundancy and optimizing data transmission within the IoT-enabled WSN environment.

To analyze the proposed scheme in terms of the fixed size chunk, total number of chunks, variable chunk size, and probability of cut-point identification with the existing state-of-the-art schemes to gauge its efficiency compared to the existing scheme.

The rest of the article is ordered as follows. Section presents a brief description of concepts and terminology, and section  investigates the literature review. Section "Sensor node (SN)" presents the system model and identification of the problem. Section  offers the proposed Data De-duplicated Scheme (DDS), which includes the CCIA to solve the identified weak points of the base schemes. Section discusses the results and study of the proposed article, and Section concludes the article.

Preliminaries

The proposed scheme involves four entities, briefly described below, before delving into the scheme itself.

Sensor node (SN)

A sensor node or an IoT device is typically employed for data collection, serving as small, resource-constrained, specialized devices designed for specific tasks. In healthcare, for instance, these devices monitor and collect patient data, including parameters such as blood pressure, heartbeat rate, and blood sugar. Due to their inherent constraints, it is crucial to use these devices efficiently to conserve their resources and prevent rapid depletion27.

Aggregator node

Many sensor nodes (SN) could be deployed in a specific area to collect data. These SNs forward their collected data to a node that aggregates the data from all the SNs, called the Aggregator Node (AN). The AN plays a crucial role in conserving the energy sources of SN devices by preventing them from transmitting over long distances27.

Fog server

The Fog Server (FS), strategically positioned at the edge where data is generated, acts as an intermediary between the Aggregator Node (AN) and the Cloud Server (CS). Additionally, it is crucial to eliminate redundant or duplicate data collected by two or more Sensor Nodes (SN) from the same source before forwarding it to a distant CS. By removing duplicate data, the FS not only decreases the size of the data to be transmitted to a remote CS but also minimizes communication overhead on the network. This improvement contributes to enhanced transmission efficiency and reduced latency, which is particularly beneficial for delay-sensitive applications28.

Cloud server

The Fog Server (FS) transmits the deduplicated data to be stored on the Cloud Server (CS). Deduplication at the FS enhances communication between the FS and the CS, especially when the distance between them is considerable28.

Literature review

This section presents the literature review on data deduplication schemes, the surveys on IoT-enabled WSNs, and an analysis of these schemes. The literature covers the fixed_sized windows and variable-sized windows for data transmission. The authors offer the “P-Dedupe: Exploiting Parallelism in Data Deduplication Systems”29, which makes use of the parallelisms of CDC-based deduplication tasks in deduplication systems. P-Dedupe first parallelizes content-defined chunking before piping the four deduplication phases using those chunk and file processing units. P-Dedupe separates the data stream into numerous segments, each performing CDC in parallel using a separate thread. To assure the parallelization of CDC’s chunking efficiency, P-Dedupe swiftly chunks and joins the two chunks at the boundaries of the neighboring segments. Using many CPU cores lowers the data deduplication ratio while boosting deduplication throughput nearly linearly. Because sensor devices have limited resources, they could not be better for IoT-enabled Wireless Sensor Network environments.

In30, the authors propose a few secure data de-duplication-based schemes on Rabin_fingerprinting. The method is realized, made on “deterministic_tags and random_tags”. In the article, deduplication enables the data to be outsourced to a storage server before the cloud. In particular, they used “variable-size block-level deduplication” by using “Rabin fingerprinting.” Limitations of this scheme are as follows: The deduplication time depends on the sliding step size of the window. The time required for block generation of this scheme is more than the fixed-size-block scheme. The data duplication ratio of the fixed-size-block scheme is better than the Rabin fingerprinting scheme. The throughput of the fixed-size-block scheme is better than the Rabin fingerprinting scheme. In31, the Asymmetric Extremum (AE) algorithm is proposed using two windows: a “variable-sized window (VSW)” set to the left of the “fixed-sized window (FSW)”. The highly valued byte is fitted in the middle of two windows. AE scans every byte of the FSW to search the cut-point. For every byte, the AE compares all the values for each byte with all bytes in the FSW. The FSW is always fitted next to every scanned byte. If the capacity of the value of the byte is more significant than all of each byte in the FSW, then it gets to the identified cut-point. The main issue in this scheme is that the size of the VSW will be zero when the leading bytes are valid for the condition as it compares each byte to the FSW, so an account of this computation cost increases.

In the article32, the authors propose a scheme based on content-defined chunking for multiple purposes. Utilizing an asymmetric chunking technique, they introduce a new algorithm called ’Rapid-Asymmetric-Maximum (RAM).’ The RAM algorithm is designed to search for a significant value byte in every Fixed-Size Window (FSW). If the byte immediately following the FSW has a more significant value than that in the FSW, this byte is identified as the maximum valued byte, facilitating the determination of a cut-point. The algorithm then proceeds to the next immediate byte. It continues until it reaches a byte with a more significant value. In this scheme, the FSW is located to the left of the Variable-Size Window (VSW). Notably, the VSW is considered less significant than the FSW. If a VSW is found at the third byte after the FSW and is larger than the most significant byte of the FSW, it indicates that a VSW may not always be larger than the FSW. It has been observed that in both AE and RAM, the FSW may be larger than the VSW. In both the schemes, the cut-point is identified based on the certain value either maximum or significant that may not be feasible in many cases which results in generating very large sized variable window that may cause high computational cost. On the contrary, the cut-point condition met at earlier stage may results in generating smaller sized chunks where the chances of duplication will also be reduced. The execution of the deduplication algorithm is completely performed, but the deduplication rate remains low, and more computational cost is consumed with poor results.

An efficient de-duplication method for the cluster de-duplication system describes the cluster de-duplication system. ARDedup is designed to manage load balancing while achieving a high data de-duplication rate with minimal communication overhead. To efficiently examine network activity patterns and identify potential breaches in real-time communication, a deep neural network-based framework is presented33. The authors suggested a model that uses an optimized CNN by adjusting the CNN’s parameters using the grey wolf optimizer (GWO) method, which helps to increase the model’s prediction accuracy for finding duplicates.34.

In the developing system of smart devices, over-utilization is a complex problem that leads to higher expenses, poor performance, and low prices because of “service level agreements (SLA)”35. A data block segmentation is performed by using a dual sliding window. Moreover, a cut-point prediction algorithm is adopted for efficient cut-point selection. No fixed chunk size is considered in this scheme. The deduplication ratio is measured by utilizing the presented calculation procedure. The discussed scheme is compared with known related algorithms by using a public dataset. The DSW algorithm performs well in terms of data ocean. The experiment results show the supremacy of the presented scheme when compared with the related schemes36.

A Dynamic Deduplication Approach for Big Data Storage explores a dynamic deduplication approach with greater efficiency. It uses content defined-chunking, where the entire file is divided into blocks, and then the hash is taken using SHA1 as an identifier for that block. CDC prevents the boundary shift problem of the static to the chunking approach.

The scheme first separates the input data into halves. It uses Rabin-based CDC to form chunks and then generates hash values of each chunk using SHA137. The weak point of the scheme is that it consumes more time than another scheme where the authors claim to transmit more data per symbol by using better transceiver architecture. To ensure the end user’s quality of experience, the scheme suggests an online joint offloading and resource allocation (JORA) system under the long-term MEC energy restriction38. It has been observed that more computation cost is required along with additional energy consumption due to hash calculation for each chunk before transmission. The scheme considers secure deduplication, which requires the additional cost of providing security.

In contrast to38, the authors of “QuickDedup” claimed to reduce the number of hashes for identifying the duplicates. It also reduces the deduplication time and related overheads. The scheme arranges the data in different classes of blocks and eliminates the duplicates so that the hashing and related operations are not performed for those blocks. Moreover, the storage requirement is also reduced, which results in the overall quick execution of the deduplication process39. In contrast to the schemes38 and39, the methods in “FastCDC” are claimed to be more efficient in terms of executing the deduplication40. In the FastCDC approach, the hash calculation judgment is improved, and it also skips the cut-points that result in sub-minimum chunks. Lastly, the scheme normalizes the chunk sizes. The scheme claims to be 10 times faster than the existing open source CDC schemes and three times faster than the AE based approaches.

A hybrid parallel learning approach41 builds two SCNs simultaneously. One uses a point-incremental algorithm, while the other uses a block-incremental approach. In42, the authors designed a static memory deduplication scheme for reducing memory storage conditions. In43, the authors created the deduplication method by altering the related features in CDC to find the Cut-points. In44, they suggested jointly looking into a challenge involving content caching and task offloading in D2D-assisted MEC networks to optimize energy consumption and task completion latency. In45, the authors presented a novel method of data encryption that combines dynamic key management (DKM) with AES-256, or Improved Elliptic Curve Cryptography (IECC). The study balances data security and efficient communications in the Internet of Vehicles.

Data redundancy is proportional to data authentication in an indirect way. The probability of authenticating specific duplicate data decreases as redundancy grows. Despite the rapid expansion of cloud computing and big data, deduplication is mandatory to reduce storage costs. Deduplication reduces the cost of cloud storage by preventing the same data from being stored numerous times in real-time. Encrypting client data in the server allows for secure deduplication. It must inspire clients to believe in the service provider46. A deduplication approach for data storage systems applies a greedy algorithm. Each segment identifies container consumption, and old containers are adjusted dynamically. Moreover, this scheme resists the storage of deduplicated chunks to enhance the restoration time. Furthermore, the greedy plus algorithm is also considered for more efficient identification of container consumption and allocation of reference containers on the global scope within a single interval. The experimental values show that this scheme enhances the restore time while maintaining the system backup performance47.

A jump-based chunking algorithm provides effective data deduplication ratio and throughput. In this scheme, the sliding window can skip a particular size of the input data stream when a jump condition is satisfied. The effect of the cut point and jump point on the chunk size is analyzed. The experiment result shows that this scheme provides efficient chunking throughput by utilizing the jump condition while maintaining the high deduplication ratio in48. The authors in49 present a novel cyberattack prediction paradigm for IoT systems. It utilizes the updated CHAID decision tree machine learning techniques and multi-class support vector machines (SVM).

The careful review of the literature explores the specific gaps identified in the base schemes31,32 as explored in the next Section. It resulted in the design of the proposed scheme to present a dependable solution that mitigates these identified gaps. A few of the limitations and gaps are also discussed as follows. It has been observed that the buffer schedule must be disrupted when small or larger chunk sizes are created and transmitted. It is a crucial issue for researchers to address effectively, as it leads to system inefficiency. It is also observed that the researchers aim to identify duplicate files only in the previously stored locations by comparing them with the most recent files37. Identifying duplicated files in the entire system is also mandatory. Moreover, the researchers focus on word frequency-based deduplication to reduce the possibility of eliminating redundant data.

Problem identification

In an IoT environment, redundant data storage in the Fog (FoG) or cloud consumes valuable storage resources. Therefore, eliminating redundant values before exchanging or storing information is crucial. Data deduplication is performed on data chunks, and various deduplication schemes with chunking algorithms are presented to enhance the effectiveness of the chunking process. In31, the Asymmetric Extremum (AE) algorithm is introduced, wherein the authors proposed the use of two types of windows: variable_sized_window (VSW) and Fixed_Sized_Window (FSW).

In the AE algorithm, the VSW is set before the FSW, and the FSW values are added and compared with the possible maximum value to iterate the loop. Two steps are then executed to compare the index with the max-position plus window size. In this method, the most significant byte is positioned at the center of the two windows. AE scans every byte of the VSW to find the cut-point where the condition is true, similar to identifying the highest value from the values in FSW.

However, a notable issue in the AE scheme is that the size of the VSW may become zero bytes when the condition is met at the first byte. Additionally, the computation cost of the AE algorithm increases as each byte in the FSW is compared. In32, the RAM scheme is introduced to address this issue by searching for a significant value byte in every FSW. Suppose a byte immediately follows the FSW and has a more substantial value than one in the Fixed-Size Extended Window (FSEW) value. This byte is used as the maximum value byte, and the cut-point is identified. The algorithm then proceeds to the next immediate byte until it determines the more significant value bytes. Notably, in this scheme, the FSW occurs to the left of the VSW, and the VSW is equal to the FSW. If found at the third byte after FSW, it is larger than the most significant byte of the FSW.

System model

The system model for smart healthcare systems is illustrated in Fig. 2. In the system paradigm, numerous patients are grouped, each with a few smart monitoring sensor devices attached to their bodies. These sensor nodes collect the patient’s health information and forward it to the Aggregator node (AN). The AN collects all information about the given region and then transmits this information to the fog server as a data stream. The fog server further sends data to the cloud server.

Moreover, the transmission of redundant values enhances the transmission cost and storage costs both at the fog server and cloud servers. Therefore, the data deduplication method is proposed to provide data deduplication at the fog and cloud servers. In this context, the fog server dedicates data and stores the received data in the local storage. Furthermore, the fog server forwards the deduplicated data to the cloud server to reduce the storage cost at the cloud server. In real-world scenarios, the patients’ data is collected on regular basis and stored at the central repositories. In actual, a number of redundant readings of the patient’s healthcare parameters including temperature, heart rate, blood pressure, oxygen level, ECG etc. may remain the same for many hours or even the entire day. It demands the use of deduplication to ensure that redundant values are not stored again and again on central repositories to enhance the storage cost massively. In real-time applications, the data is uploading continuously after every certain seconds as per the level of criticality of the patient. The proposed solution helps to reduce the data generation and storage for the healthcare applications in the real-world.

We considered the patients within their vicinity like home, office or hospital within a certain range of mobility from the collector node that takes reading for the healthcare sensors. Our solution can adopt the mobility of the patient if the collector node like cellular phone or any device is also moving with that person. The aggregated data is transmitted to the Fog server via internet connectivity. It ensures to handle the changing conditions due to the mobility of the person.Figure 2 Proposed FoG-based data collection and storage in IoT.

Proposed data deduplicated scheme (DDS)

This Section presents the proposed Data De-duplicated Scheme (DDS) that efficiently manages the data chunks by identifying the VSW of sufficient size. The rationale of this work is to reduce the cost raised by repeatedly transmitting the redundant data values. This work mitigates the duplication to reduce the communication cost by the aggregator nodes and also reduce the storage cost at central repositories. The proposed approach consists of four phases: Data collection, Data aggregation, Data deduplication, and Data storage. According to the deployment model, the flow operations for data processing are illustrated in Fig. 3.Figure 3 The flow of data processing in the proposed scheme.

Medical sensor nodes continuously monitor the patient’s information during the data processing flow and transmit it to the Aggregator Node for further processing. The Aggregator Node gathers all relevant patient data. Subsequently, this information is transmitted as a data stream to the Fog Server. The Fog Server executes the CCIA algorithm to establish a dominant chunk size, identifies duplicate data, and initiates the deduplication process for that data. The deduplicated data is then stored locally in the Fog Server. Upon completion of the deduplication process, the de-duplicated data is sent to the Cloud Server. Table 1 shows a used notation of the article.Table 1 List of Notations.

Symbol	Description	Symbol	Description	
L	Input string length	ACH	2 dimensional array	
ep	End Point	sp	Starting Point	
WFS	Fixed Window Size	WVS	Variable Window Size	
Minoff	Minimum offset	Uoff	Upper Offset	
CPI	Cut-Point- Identification			

Controlled cut-point identification algorithm

A Controlled Cut-point Identification Algorithm (CCIA) is proposed in this scheme to identify a suitable cut-point in a window. The CCIA uses a 2-dimensional array to determine the controlled cut-point. The proposed CCIA provides cut-point identification while creating chunks. CCIA has the first window as fixed-sized, while the rest are variable-sized. The length of the chunk is based on the variable window size. In CCIA, the string length is entered in a 2-dimensional array. The pseudo-code of the proposed CCIA is illustrated in algorithm 1.

Algorithm 1 Controlled Cut-point Identification Algorithm (CCIA)

The steps of algorithm 1 are elaborated as follows.

In step 1, parameters such as i, sp, and IN are initialized. From steps 2 to 4, the size of the fixed window WFS is set by assigning the value of i to the zero index position of the 2-dimensional array to create the FSW. Then, the ep is set (ep)=i+WFS to the value at the end of the FSW. In step 5, the index i is first assigned position 1 to create the VSW in a 2-dimensional array. Steps 6 to 8, in the case of WVS (Variable window size), sp is set as the for WVS at the end of the fixed window size. For variable size window, set sp=ep+1, IN=IN+1, i=1, WVS=0. Assign the ep and WFS to the sp of the VSW. After that, assign the random value selected from the WFS/2 and Uoff to the offset of the 2-dimensional array.

Step 9 to 10, calculate the Minoff by dividing the FSW with (WFS/2) and also calculate Uoff by subtracting the 2WFS from the Maxchunk_length, and assign it to offset, also assign start and offset to the endpoint array. Moreover, it assigns sp and the offset of random length between Minoff and Uoff to the ep of the 2-dimensional array and then assigns this ep to index[1] position of the 2-dimensional array. Step 11 gives the ep subtracted by the FSW index position to WVS of the index position of a 2-dimensional array. Steps 12 to 13 assign the ep+1 to the index [i]. After that, assign this index[i] to the IN, which is initially set to zero, and repeat step 2 to the end for the length of the string, add offset values in the starting point, and set a cut point at the endpoint to form a chunk. Moreover, the CCIA shows that the VSW must dominate the existing scheme. CCIA provides an average length of chunks and avoids extra small and large chunks. The proposed method is illustrated through an example shown in Fig 4 (a),(b), and (c). In Fig. 4, the chunking and cut-point identification are shown for (a) AE31, (b) RAM32, and (c) the proposed CCIA.Figure 4 The chunking and cut-point identification for (a) AE, (b) RAM, and (c) CCIA scheme.

In the example of AE31, it is clearly shown that the size of the VSW is smaller and may be zero if the byte of EA appears at the beginning of the data streams. This results in smaller chunk sizes and low chances of duplicates. In the example of the RAM32, the VSW is smaller, and the cut-point byte will occur next to the ESW. The smaller the VSW, the smaller the chances of duplication. The example shows that the proposed scheme is better at comparing the sizes of the VSW. The proposed CCIA algorithm uses a 2-dimensional array to identify a controlled cut-point, which should be dominant and more significant than the FSW. According to this argument, the VSW of the proposed algorithm is larger than the existing schemes. The CCIA algorithm shows that the size of chunks will be increased, leading to higher chances of duplication than the existing schemes. In the worst case, the proposed algorithm’s average chunk size is larger than the existing schemes.

In the data processing flow, all the sensor nodes monitor the patient’s information and send it to the AN for further processing. The AN collects all the information of the corresponding patients. After gathering information, the information is sent as a data stream to the fog server. The fog server identifies duplicate data and performs the deduplication process using the CCIA algorithm to create a dominant chunk size. The server then stores de-duplicated data in the local storage. After the data deduplication, the fog server sends the de-duplicated data to the Cloud Server.

Results and analysis

The experiments are conducted using a testbed which is using the “Windows Communication Foundation (WCF)” services set up on “Windows Azure Cloud.” together with C-Sharp (C#) and ASP.net. The same medical data streams are used to analyze the performance of the CCIA scheme against AE31 and RAM32. An Android app is designed to enhance the data streams by storing more medical information in the database as per the sensing limit.

This Section analyzed that the CCIA scheme provides effective results of the minimum, maximum, and average chunk sizes. Moreover, we also compare the CCIA scheme with the total number of chunks, fixed and variable chunk size, and probability of cut point identification. The computational overhead of the proposed scheme is less than the existing approaches. The proposed scheme adds a value to the index, a jump more significant than FSW. It requires less cost than the existing schemes, which compare each byte of the input text. The simulation parameters are provided in Table 2Table 2 Simulation Parameters.

Symbol	Description	
Target area	1100 × 1100 m	
Multiple devices	30–200	
Group area	400 m	
Mobility range	50 m	
No of data attributes	20–80	
Nodes per Group	10–40 nodes	
Original Unit Data Size	50–500 bytes	
Maximum chunk threshold	2048 bytes	
Minimum chunk threshold	512 bytes	
Variable window length	800 bytes	

.

In the case of existing schemes, the AE scheme has the first variable size window and the second fixed window size to achieve a large chunk size. It does not ensure that the cut-point can be identified to produce a tiny chunk earlier to have more chunks or very late to form a huge chunk, i.e., fewer chunks, which will not result in optimized deduplication. The RAM scheme provides a variable-size double window and attempts to find the cut point after the end of the fixed window index. It requires more computation cost and may miss the criteria to meet the threshold. This results in reaching an extra large chunk size and reduces the total number of chunks that require more computational cost for deduplication. The proposed CCIA achieved a sufficient number of chunks to achieve better deduplication. Fig. 5 illustrates that the CCIA, RAM, and AE are analyzed in terms of the total number of chunks at the fog server against six different data_streams. The CCIA, RAM, and AE generate 184930 chunks, 196342 chunks, and 205132 chunks in the case of data_stream-1. Moreover, The CCIA, RAM, and AE generate 389273 chunks, 419331 chunks, and 448740 chunks in the case of data_stream-2. The proposed scheme dominates the chunk size. CCIA avoids the generation of extra small chunk sizes.Figure 5 Total Number of Chunks.

The average chunk size is illustrated in Fig. 6a. The average AE, RAM, and CCIA chunk size is examined against the different data_streams. The AE and RAM schemes do not ensure a sufficient number of chunks in large sizes and may suffer from tiny or zero-sized chunks, which also reduces the average size. The proposed CCIA provides that the ending index of the variable-sized window is found after a certain threshold with the upper side of offset to ensure all the chunks are larger than a threshold, which confirms the better average chunk size. In data_stream-1 and 3, the average chunk size is 1132 bytes, 1054 bytes, and 1023 bytes for AE, RAM, and CCIA schemes, and for AE, RAM, and CCIA schemes is 1462 bytes, 1261 bytes, and 1236 bytes. Set each larger than 3/4 of the max-chunk to achieve an average chunk size. Fig. 5 and Fig. 6a compare AE, RAM, and Our Proposed scheme. The total number of chunks and the average chunk number are presented in Table 3. The data_stream presented in Table 3 demonstrates a comparison of the proposed CCIA’s AE and RAM approaches with the proposed CCIA’s total number of chunks at the fog server versus six different data_streams. The table shows that the suggested plan prevails in terms of chunk sizes. CCIA avoids the generation of extra small chunk sizes. Therefore, it increases the overall count by generating additional chunks. Also, the average chunk number is illustrated in Table 3, where the average chunk size of AE, RAM, and CCIA is examined against different data_streams.Table 3 Total Number of Chunks and Average Chunks Number.

Data_Stream	Total Number of Chunks	Average Chunk Number	
AE31	RAM32	CCIA	AE31	RAM32	CCIA	
D1	205132	196342	184930	1132	1054	1023	
D2	448740	419331	389273	1673	1365	1297	
D3	314234	303302	282304	1462	1261	1236	
D4	96354	88736	82692	1387	1183	1131	
D5	263542	255394	239832	1359	1237	1213	
D6	338736	334702	302304	1435	1262	1242	

The chunk size plays an essential role in analyzing the influence of deduplication. In Fig. 6b, it is analyzed that more minimum-sized chunks are generated in the case of AE and RAM. There is no check on the size of the chunk. Therefore, a tiny chunk can be generated if the condition occurring at a cut point is met. The cut-point may be identified exactly after one byte, the first byte of the variable-sized window, which starts after the last byte of the fixed-sized window. Therefore, the chances of the small-sized VSW are relatively high. The results show that AE, RAM, and CCIA provide minimum chunk sizes of 84 bytes, 138 bytes, 642 bytes for data_stream-1 and 315 bytes, 379 bytes, 773 bytes for data_stream-2, respectively. CCIA achieves adequate chunk size by setting a minimum threshold limit value because extra small_size chunks decrease the performance of data deduplication and redundant data identification.Figure 6 At the fog server, the data stream is plotted against a Average chunk number and b Minimum chunk size.

On the other hand, extra-large size chunks enhance both computational and storage costs. Therefore, CCIA avoids extra-large chunk size by setting a maximum threshold limit. In Fig. 7, the maximum chunk sizes of the proposed CCIA scheme are comparatively analyzed with the related schemes on several data_streams. Results show the dominance of the CCIA scheme in both minimum and maximum chunk sizes. There is a comparison between the smallest and largest chunk size in Table 4.Figure 7 Maximum Chunk Size.

Table 4 Minimum Chunk Size And Maximum Chunk Size.

Data_Stream	Minimum Chunks Size	Maximum Chunks Size	
AE31	RAM32	CCIA	AE31	RAM32	CCIA	
D1	84	138	642	3053	1602	1534	
D2	315	379	773	5701	1953	1953	
D3	242	201	662	7094	1701	1650	
D4	202	232	587	6462	1806	1783	
D5	272	281	589	5956	1791	1740	
D6	253	251	635	6461	1854	1871	

A comparison of the failure rate to identify the Probability of the Cut-point is presented in Table 5. The table demonstrates how the proposed CCIA is better than other current methods, AE and RAM, in the failure probability of Cut-point Identification. The probability is calculated in terms of the computational cost. The results show that AE provides a higher failure probability than RAM and CCIA. The proposed CCIA is better than the existing schemes.Table 5 Failure Probability of the Cut-point Identification.

CPI	AE31	RAM32	CCIA	
2	37	26	11	
4	23	18	8	
6	19	11	4	
8	8	5	2	

The CCIA scheme is compared with different deduplication-based schemes regarding variable window size. Fig. 8 shows the variation in varying window sizes like AE, RAM, and CCIA provide varying window sizes of 764, 770, and 800, respectively.Figure 8 Variable-Window-Size.

In Fig 9, we analyze the failure probability of cut-point identification in the chunking process. It has been observed for AE and RAM that the cut-point identification condition may fail even when a very large-sized variable window is processed. The window may be discarded in such cases, resulting in chunking failure. In the case of the proposed CCIA, the chunk is decided based on the delimiter, which is the mandatory part for the aggregated message to differentiate the readings of one patient from another in the healthcare scenario. It identifies the cut-point with a very high probability. Results show that AE has a higher failure probability than RAM and CCIA.Figure 9 Probability of Cut-point Identification.

Finally, based on the above results, it is evident that the proposed scheme outperforms its competitors. This superiority is primarily due to the scheme’s enhanced ability to eliminate duplicate data chunks. The scheme improves space efficiency and reduces computational and communication overhead by achieving this. Consequently, this reduction directly contributes to the scalability of the scheme, enabling it to support a more significant number of devices and scale more effectively.

Conclusion

In conclusion, this article introduced a data duplication scheme designed to mitigate storage and communication overhead. By addressing the transmission and storage of duplicate data from testbed cloud repositories, the proposed CCIA algorithm efficiently identifies controlled cut-points, thereby creating variable-sized windows and avoiding redundant data storage and communication costs. Unlike existing schemes where the variable-sized window could be reduced to zero bytes upon fulfilling a condition after the fixed-size window, the proposed approach strategically places the controlled cut-point using the CCIA algorithm. The validation of our scheme on a test bed executing WCF services on a Windows Azure cloud, implemented with ASP.net and C#, demonstrated the effectiveness of the CCIA algorithm at the fog server. Our results showcased the superiority of CCIA in various metrics, including the average number of chunks, the total number of chunks, variable chunk size, the minimum and maximum size of chunks, and the probability of failure for cut point identification.

The limitation of the scheme may be its simulation environment. Implementation on a real testbed or practical environment may affect the results. However, it may not significantly affect the general analysis derived from the current results. Its primary focus is identifying duplicate blocks by examining individual data transmissions and looking ahead to explore the word frequency-based de-duplication methods to minimize the risk of eliminating redundant data. To improve the effectiveness of removing redundant data, we will investigate word frequency-based de-duplication techniques in the future. This research contributes to the ongoing efforts to optimize data storage and communication efficiency in cloud-based environments.

Acknowledgements

This research work was supported by the National Research Foundation of Korea (NRF) under the Brain Pool Program (Grant No. RS-2023-00220285). This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT)(No. RS-2024-00423362). Any correspondence related to this paper should be addressed to Do-Hyeun Kim.

Author contributions

All authors contributed equally to this article.

Data availibility

All data generated or analyzed during this study are included in this article.

Competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Nyangaresi VO Privacy preserving three-factor authentication protocol for secure message forwarding in wireless body area networks Ad. Hoc. Netw. 2023 142 103117 10.1016/j.adhoc.2023.103117
Nyangaresi, V. O. Privacy preserving three-factor authentication protocol for secure message forwarding in wireless body area networks. Ad. Hoc. Netw. 142, 103117. 10.1016/j.adhoc.2023.103117 (2023).10.1016/j.adhoc.2023.103117
2. Badshah A Towards smart education through internet of things: A survey ACM Comput. Surv. 2023 10.1145/3610401
Badshah, A. et al. Towards smart education through internet of things: A survey. ACM Comput. Surv.[SPACE]10.1145/3610401 (2023).10.1145/3610401
3. Chen M Szu HF Lin HY Liu Y Chan HY Wang Y Zhao Y Zhang G Yao JD Li WJ Phase-based quantification of sports performance metrics using a smart IoT sensor IEEE Internet Things J. 2023 10 18 15900 11 10.1109/JIOT.2023.3266351
Chen, M. et al. Phase-based quantification of sports performance metrics using a smart IoT sensor. IEEE Internet Things J. 10(18), 15900–11 (2023).10.1109/JIOT.2023.3266351
4. Dahooie JH Mohammadian A Qorbani AR Daim T A portfolio selection of internet of things (iots) applications for the sustainable urban transportation: A novel hybrid multi criteria decision making approach Technol. Soc. 2023 75 102366 10.1016/j.techsoc.2023.102366
Dahooie, J. H., Mohammadian, A., Qorbani, A. R. & Daim, T. A portfolio selection of internet of things (iots) applications for the sustainable urban transportation: A novel hybrid multi criteria decision making approach. Technol. Soc. 75, 102366 (2023).10.1016/j.techsoc.2023.102366
5. Zeb H Zero energy IoT devices in smart cities using RF energy harvesting Electronics 2023 10.3390/electronics12010148
Zeb, H. et al. Zero energy IoT devices in smart cities using RF energy harvesting. Electronics[SPACE]10.3390/electronics12010148 (2023).10.3390/electronics12010148
6. Smys S A survey on internet of things (iot) based smart systems J. ISMAC 2020 2 181 189 10.36548/jismac.2020.4.001
Smys, S. A survey on internet of things (iot) based smart systems. J. ISMAC 2, 181–189 (2020).10.36548/jismac.2020.4.001
7. Ghani A Spread spectrum based energy efficient collaborative communication in wireless sensor networks PLoS ONE 2016 11 01 24 10.1371/journal.pone.0159069
Ghani, A. et al. Spread spectrum based energy efficient collaborative communication in wireless sensor networks. PLoS ONE 11, 01–24 (2016).10.1371/journal.pone.0159069
8. Ali Z A clogging resistant secure authentication scheme for fog computing services Comput. Netw. 2021 185 107731 10.1016/j.comnet.2020.107731
Ali, Z. et al. A clogging resistant secure authentication scheme for fog computing services. Comput. Netw. 185, 107731. 10.1016/j.comnet.2020.107731 (2021).10.1016/j.comnet.2020.107731
9. Li X Lu Y Fu X Qi Y Building the internet of things platform for smart maternal healthcare services with wearable devices and cloud computing Futur. Gener. Comput. Syst. 2021 118 282 296 10.1016/j.future.2021.01.016
Li, X., Lu, Y., Fu, X. & Qi, Y. Building the internet of things platform for smart maternal healthcare services with wearable devices and cloud computing. Futur. Gener. Comput. Syst. 118, 282–296 (2021).10.1016/j.future.2021.01.016
10. Ghani A Naqvi SHA Ilyas MU Khan MK Hassan A Energy efficiency in multipath rayleigh faded wireless sensor networks using collaborative communication IEEE Access 2019 7 26558 26570 10.1109/ACCESS.2019.2898565
Ghani, A., Naqvi, S. H. A., Ilyas, M. U., Khan, M. K. & Hassan, A. Energy efficiency in multipath rayleigh faded wireless sensor networks using collaborative communication. IEEE Access 7, 26558–26570 (2019).10.1109/ACCESS.2019.2898565
11. Sun Z Zhao G Xing X Encp: A new energy-efficient nonlinear coverage control protocol in mobile sensor networks EURASIP J. Wirel. Commun. Netw. 2018 2018 1 15 10.1186/s13638-018-1023-7
Sun, Z., Zhao, G. & Xing, X. Encp: A new energy-efficient nonlinear coverage control protocol in mobile sensor networks. EURASIP J. Wirel. Commun. Netw. 2018, 1–15 (2018).10.1186/s13638-018-1023-7
12. Yi L A novel block encryption algorithm based on chaotic s-box for wireless sensor network IEEE Access 2019 7 53079 53090 10.1109/ACCESS.2019.2911395
Yi, L. et al. A novel block encryption algorithm based on chaotic s-box for wireless sensor network. IEEE Access 7, 53079–53090 (2019).10.1109/ACCESS.2019.2911395
13. Khashan OA Hybrid lightweight proxy re-encryption scheme for secure fog-to-things environment IEEE Access 2020 8 66878 66887 10.1109/ACCESS.2020.2984317
Khashan, O. A. Hybrid lightweight proxy re-encryption scheme for secure fog-to-things environment. IEEE Access 8, 66878–66887 (2020).10.1109/ACCESS.2020.2984317
14. Chen M Szu HF Lin HY Liu Y Chan HY Wang Y Zhao Y Zhang G Yao JD Li WJ Phase-Based Quantification of Sports Performance Metrics Using a Smart IoT Sensor IEEE Internet Things J. 2023 10 18 15900 11 10.1109/JIOT.2023.3266351
Chen, M. et al. Phase-Based Quantification of Sports Performance Metrics Using a Smart IoT Sensor. IEEE Internet Things J. 10(18), 15900–11 (2023).10.1109/JIOT.2023.3266351
15. Sharma N Prasad AK Kakulapati V Data deduplication techniques for big data storage systems Int. J. Innov. Technol. Explor. Eng. 2019 8 1145 1150 10.35940/ijitee.J9129.0881019
Sharma, N., Prasad, A. K. & Kakulapati, V. Data deduplication techniques for big data storage systems. Int. J. Innov. Technol. Explor. Eng. 8, 1145–1150 (2019).10.35940/ijitee.J9129.0881019
16. Chen M Szu HF Lin HY Liu Y Chan HY Wang Y Zhao Y Zhang G Yao JD Li WJ Phase-Based Quantification of Sports Performance Metrics Using a Smart IoT Sensor IEEE Internet Things J. 2023 10 18 15900 11 10.1109/JIOT.2023.3266351
Chen, M. et al. Phase-Based Quantification of Sports Performance Metrics Using a Smart IoT Sensor. IEEE Internet Things J. 10(18), 15900–11 (2023).10.1109/JIOT.2023.3266351
17. Hovhannisyan H Qi W Lu K Yang R Wang J Whispers in the cloud storage: A novel cross-user deduplication-based covert channel design Peer-to-Peer Netw. Appl. 2018 11 277 286 10.1007/s12083-016-0483-y
Hovhannisyan, H., Qi, W., Lu, K., Yang, R. & Wang, J. Whispers in the cloud storage: A novel cross-user deduplication-based covert channel design. Peer-to-Peer Netw. Appl. 11, 277–286 (2018).10.1007/s12083-016-0483-y
18. Periasamy JK Latha B An enhanced secure content de-duplication identification and prevention (ESCDIP) algorithm in cloud environment Neural Comput. Appl. 2020 32 2 485 94 10.1007/s00521-019-04060-9
Periasamy, J. K. & Latha, B. An enhanced secure content de-duplication identification and prevention (ESCDIP) algorithm in cloud environment. Neural Comput. Appl. 32(2), 485–94 (2020).10.1007/s00521-019-04060-9
19. Periasamy JK Latha B An enhanced secure content de-duplication identification and prevention (ESCDIP) algorithm in cloud environment Neural Comput. Appl. 2020 32 2 485 94 10.1007/s00521-019-04060-9
Periasamy, J. K. & Latha, B. An enhanced secure content de-duplication identification and prevention (ESCDIP) algorithm in cloud environment. Neural Comput. Appl. 32(2), 485–94 (2020).10.1007/s00521-019-04060-9
20. Cao B Sun Z Zhang J Gu Y Resource allocation in 5g iov architecture based on sdn and fog-cloud computing IEEE Trans. Intell. Transp. Syst. 2021 22 3832 3840 10.1109/TITS.2020.3048844
Cao, B., Sun, Z., Zhang, J. & Gu, Y. Resource allocation in 5g iov architecture based on sdn and fog-cloud computing. IEEE Trans. Intell. Transp. Syst. 22, 3832–3840 (2021).10.1109/TITS.2020.3048844
21. Sadri AA Rahmani AM Saberikamarposhti M Hosseinzadeh M Fog data management: A vision, challenges, and future directions J. Netw. Comput. Appl. 2021 174 102882 10.1016/j.jnca.2020.102882
Sadri, A. A., Rahmani, A. M., Saberikamarposhti, M. & Hosseinzadeh, M. Fog data management: A vision, challenges, and future directions. J. Netw. Comput. Appl. 174, 102882 (2021).10.1016/j.jnca.2020.102882
22. Chaudhry SA Correcting, “palk: Password-based anonymous lightweight key agreement framework for smart grid” Int. J. Electrical Power Energy Syst. 2021 125 106529 10.1016/j.ijepes.2020.106529
Chaudhry, S. A. Correcting, “palk: Password-based anonymous lightweight key agreement framework for smart grid’’. Int. J. Electrical Power Energy Syst. 125, 106529 (2021).10.1016/j.ijepes.2020.106529
23. Begum BR Chitra P Ecc-crt: An elliptical curve cryptographic encryption and chinese remainder theorem based deduplication in cloud Wireless Pers. Commun. 2021 116 1683 1702 10.1007/s11277-020-07756-7
Begum, B. R. & Chitra, P. Ecc-crt: An elliptical curve cryptographic encryption and chinese remainder theorem based deduplication in cloud. Wireless Pers. Commun. 116, 1683–1702 (2021).10.1007/s11277-020-07756-7
24. Wang X Zhou Q Tong J V-matrix-based scalable data aggregation scheme in wsn IEEE Access 2019 7 56081 56094 10.1109/ACCESS.2019.2913396
Wang, X., Zhou, Q. & Tong, J. V-matrix-based scalable data aggregation scheme in wsn. IEEE Access 7, 56081–56094 (2019).10.1109/ACCESS.2019.2913396
25. Chaudhry SA Rotating behind privacy: An improved lightweight authentication scheme for cloud-based iot environment ACM Trans. Internet Technol. 2021 10.1145/3425707
Chaudhry, S. A. et al. Rotating behind privacy: An improved lightweight authentication scheme for cloud-based iot environment. ACM Trans. Internet Technol.[SPACE]10.1145/3425707 (2021).10.1145/3425707
26. Rahman HU Improving network efficiency in wireless body area networks using dual forwarder selection technique Pers. Ubiquit. Comput. 2022 26 11 24 10.1007/s00779-021-01539-y
Rahman, H. U. et al. Improving network efficiency in wireless body area networks using dual forwarder selection technique. Pers. Ubiquit. Comput. 26, 11–24 (2022).10.1007/s00779-021-01539-y
27. Rabie OBJ A full privacy-preserving distributed batch-based certificate-less aggregate signature authentication scheme for healthcare wearable wireless medical sensor networks (hwmsns) Int. J. Inf. Secur. 2024 23 51 80 10.1007/s10207-023-00748-1
Rabie, O. B. J. et al. A full privacy-preserving distributed batch-based certificate-less aggregate signature authentication scheme for healthcare wearable wireless medical sensor networks (hwmsns). Int. J. Inf. Secur. 23, 51–80 (2024).10.1007/s10207-023-00748-1
28. Asghari A Sohrabi MK Server placement in mobile cloud computing: A comprehensive survey for edge computing, fog computing and cloudlet Comput. Sci. Rev. 2024 51 100616 10.1016/j.cosrev.2023.100616
Asghari, A. & Sohrabi, M. K. Server placement in mobile cloud computing: A comprehensive survey for edge computing, fog computing and cloudlet. Comput. Sci. Rev. 51, 100616. 10.1016/j.cosrev.2023.100616 (2024).10.1016/j.cosrev.2023.100616
29. Xia W Accelerating content-defined-chunking based data deduplication by exploiting parallelism Futur. Gener. Comput. Syst. 2019 98 406 418 10.1016/j.future.2019.02.008
Xia, W. et al. Accelerating content-defined-chunking based data deduplication by exploiting parallelism. Futur. Gener. Comput. Syst. 98, 406–418 (2019).10.1016/j.future.2019.02.008
30. Zhang Y Su H Yang M Zheng D Ren F Zhao Q Secure deduplication based on Rabin fingerprinting over wireless sensing data in cloud computing Secur. Commun. Netw. 2018 2018 1 9081814
Zhang, Y. et al. Secure deduplication based on Rabin fingerprinting over wireless sensing data in cloud computing. Secur. Commun. Netw. 2018(1), 9081814 (2018).
31. Zhang, Y. et al. Ae: An asymmetric extremum content defined chunking algorithm for fast and bandwidth-efficient data deduplication. In 2015 IEEE Conference on Computer Communications (INFOCOM), 1337–1345 (IEEE, 2015).
32. Widodo RN Lim H Atiquzzaman M A new content-defined chunking algorithm for data deduplication in cloud storage Futur. Gener. Comput. Syst. 2017 71 145 156 10.1016/j.future.2017.02.013
Widodo, R. N., Lim, H. & Atiquzzaman, M. A new content-defined chunking algorithm for data deduplication in cloud storage. Futur. Gener. Comput. Syst. 71, 145–156 (2017).10.1016/j.future.2017.02.013
33. Lilhore UK Dalal S Simaiya S A cognitive security framework for detecting intrusions in iot and 5g utilizing deep learning Comput. Secur. 2024 136 103560 10.1016/j.cose.2023.103560
Lilhore, U. K., Dalal, S. & Simaiya, S. A cognitive security framework for detecting intrusions in iot and 5g utilizing deep learning. Comput. Secur. 136, 103560 (2024).10.1016/j.cose.2023.103560
34. Lilhore UK Hidm: Hybrid intrusion detection model for industry 4.0 networks using an optimized cnn-lstm with transfer learning Sensors 2023 23 7856 10.3390/s23187856 37765912
Lilhore, U. K. et al. Hidm: Hybrid intrusion detection model for industry 4.0 networks using an optimized cnn-lstm with transfer learning. Sensors 23, 7856 (2023).37765912 10.3390/s23187856
35. Badshah A Ghani A Irshad A Naqvi H Kumari S Smart workload migration on external cloud service providers to minimize delay, running time, and transfer cost Int. J. Commun Syst. 2021 34 e4686 10.1002/dac.4686
Badshah, A., Ghani, A., Irshad, A., Naqvi, H. & Kumari, S. Smart workload migration on external cloud service providers to minimize delay, running time, and transfer cost. Int. J. Commun Syst. 34, e4686 (2021).10.1002/dac.4686
36. Guo S Mao X Sun M Wang S Double sliding window chunking algorithm for data deduplication in ocean observation IEEE Access 2023 11 70470 70481 10.1109/ACCESS.2023.3276785
Guo, S., Mao, X., Sun, M. & Wang, S. Double sliding window chunking algorithm for data deduplication in ocean observation. IEEE Access 11, 70470–70481. 10.1109/ACCESS.2023.3276785 (2023).10.1109/ACCESS.2023.3276785
37. Nagwanshi KK File chunking approaches 2020 Data Deduplication Approaches Concepts, Strategies, and Challenges
Nagwanshi, K. K. File chunking approaches (Concepts, Strategies, and Challenges, Data Deduplication Approaches, 2020).
38. Jiang, H., Dai, X., Xiao, Z. & Iyengar, A. K. Joint task offloading and resource allocation for energy-constrained mobile edge computing. IEEE Transactions on Mobile Computing (2022).
39. Saharan S Quickdedup: Efficient vm deduplication in cloud computing environments J. Parallel Distributed Comput. 2020 139 18 31 10.1016/j.jpdc.2020.01.002
Saharan, S. et al. Quickdedup: Efficient vm deduplication in cloud computing environments. J. Parallel Distributed Comput. 139, 18–31 (2020).10.1016/j.jpdc.2020.01.002
40. Xia W The design of fast content-defined chunking for data deduplication based storage systems IEEE Trans. Parallel Distrib. Syst. 2020 31 2017 2031 10.1109/TPDS.2020.2984632
Xia, W. et al. The design of fast content-defined chunking for data deduplication based storage systems. IEEE Trans. Parallel Distrib. Syst. 31, 2017–2031 (2020).10.1109/TPDS.2020.2984632
41. Dai W Zhou X Li D Zhu S Wang X Hybrid parallel stochastic configuration networks for industrial data analytics IEEE Trans. Industr. Inf. 2021 18 2331 2341 10.1109/TII.2021.3096840
Dai, W., Zhou, X., Li, D., Zhu, S. & Wang, X. Hybrid parallel stochastic configuration networks for industrial data analytics. IEEE Trans. Industr. Inf. 18, 2331–2341 (2021).10.1109/TII.2021.3096840
42. Qi S Lu Y Wei W Chen X Efficient data access control with fine-grained data protection in cloud-assisted iiot IEEE Internet Things J. 2020 8 2886 2899 10.1109/JIOT.2020.3020979
Qi, S., Lu, Y., Wei, W. & Chen, X. Efficient data access control with fine-grained data protection in cloud-assisted iiot. IEEE Internet Things J. 8, 2886–2899 (2020).10.1109/JIOT.2020.3020979
43. Saeed ASM George LE Data deduplication system based on content-defined chunking using bytes pair frequency occurrence Symmetry 2020 12 1841 10.3390/sym12111841
Saeed, A. S. M. & George, L. E. Data deduplication system based on content-defined chunking using bytes pair frequency occurrence. Symmetry 12, 1841 (2020).10.3390/sym12111841
44. Xiao, Z. et al. Multi-objective parallel task offloading and content caching in d2d-aided mec networks. IEEE Transactions on Mobile Computing (2022).
45. Lilhore, U. K. et al. Secure wsn architecture utilizing hybrid encryption with dkm to ensure consistent iov communication. Wireless Personal Communications 1–29 (2024).
46. Venkatachalam K Prabu P Almutairi A Abouhawwash M Secure biometric authentication with de-duplication on distributed cloud storage PeerJ Comput. Sci. 2021 7 e569 10.7717/peerj-cs.569 34401472
Venkatachalam, K. et al. Secure biometric authentication with de-duplication on distributed cloud storage. PeerJ Comput. Sci. 7, e569 (2021).34401472 10.7717/peerj-cs.569
47. Lin L Deng Y Zhou Y Zhu Y Inde: An inline data deduplication approach via adaptive detection of valid container utilization ACM Trans. Storage 2023 10.1145/3568426
Lin, L., Deng, Y., Zhou, Y. & Zhu, Y. Inde: An inline data deduplication approach via adaptive detection of valid container utilization. ACM Trans. Storage[SPACE]10.1145/3568426 (2023).10.1145/3568426
48. Jin X Accelerating content-defined chunking for data deduplication based on speculative jump IEEE Trans. Parallel Distrib. Syst. 2023 34 2568 2579 10.1109/TPDS.2023.3290770
Jin, X. et al. Accelerating content-defined chunking for data deduplication based on speculative jump. IEEE Trans. Parallel Distrib. Syst. 34, 2568–2579. 10.1109/TPDS.2023.3290770 (2023).10.1109/TPDS.2023.3290770
49. Dalal S Next-generation cyber attack prediction for iot systems: leveraging multi-class svm and optimized chaid decision tree J. Cloud Comput. 2023 12 137 10.1186/s13677-023-00517-4
Dalal, S. et al. Next-generation cyber attack prediction for iot systems: leveraging multi-class svm and optimized chaid decision tree. J. Cloud Comput. 12, 137 (2023).10.1186/s13677-023-00517-4
