E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.

1051/e3sconf/202343001097
ICMPC 2023

Secure and Sustainable Decentralized Cloud

Using IPFS

P.K. Abhilash 1*, Chidananda K1, K Sandeep1, Nikhil Reddy Molaka1, Yogesh Kumar
Awasthi 3, Samanthakurthi Rajabhishek1
1Department of Information Technology, GRIET, Bachupally, Hyderabad, Telangana.
3School of Applied and Life Sciences, Uttaranchal university, Dehradun-248007, India.

Abstract. Our paper focuses on the implementation of a

decentralized cloud storage system using the IPFS (InterPlanetary
File System) technology. The project aims to provide an alternative
to centralized cloud storage by decentralization, enhanced security,
and user control over data which is a more sustainable solution than
the existing system. In a centralized cloud storage system, data is
stored on a single server controlled by a service provider. This
centralized approach poses potential risks such as single points of
failure, data breaches, and reliance on the trustworthiness of the
provider. In contrast, our decentralized cloud storage system
distributes data across a network of interconnected nodes. Each node
stores a portion of the data, ensuring redundancy and availability. By
leveraging IPFS, which utilizes a distributed hash table and content-
addressed storage, our system ensures data integrity, fault tolerance,
and censorship resistance. Users of our system have greater control
over their data. They can participate in the network by contributing
their storage resources and accessing their files securely. The project
implementation includes features such as file upload, download,
retrieval based on IPFS hashes, and user authentication. It also
provides functionalities like displaying connected peers, managing
file metadata, and enabling file deletion. Overall, our project
showcases the potential of decentralized cloud storage, exemplifying
its benefits and illustrating how it can be implemented using IPFS
technology.

1 INTRODUCTION
The secure decentralized cloud project aims to leverage the InterPlanetary File System (IPFS)
to create a robust and privacy-focused cloud storage solution. Traditional cloud storage relies
on centralized servers, which can present vulnerabilities and concerns regarding data privacy

* Corresponding author: [email protected],

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative
Commons Attribution License 4.0 (https://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

and security. In contrast, IPFS offers a decentralized and distributed approach to storing and
retrieving files. The motivation behind this paper lies in addressing the limitations of
centralized cloud storage systems. By adopting a decentralized architecture, the project aims
to enhance data privacy, reduce reliance on single points of failure, and promote a more
resilient cloud storage solution. With the increasing concerns surrounding data breaches,
surveillance, and censorship, a secure decentralized cloud using IPFS provides an alternative
approach that aligns with the principles of privacy, security, sustainability and data
ownership. By leveraging the peer-to- peer network of IPFS, users can have greater control
over their data, ensuring that files are stored across multiple peers rather than relying on a
single centralized server. The significance of this project lies in empowering individuals and
organizations to have ownership and control over their data. By utilizing IPFS, the project
enables users to securely upload, store, and retrieve files while maintaining data privacy.
Additionally, the integration of smart contracts and cryptocurrency payments provides a
mechanism for incentivizing peers who store users' files, further enhancing the decentralized
ecosystem. Overall, the secure decentralized cloud project serves as a step towards a more
secure, private, and resilient cloud storage solution, providing users with increased control
and ownership of their data in an era of growing concerns about data privacy and security.

Machine Learning Algorithms:

Machines are used nowadays in every part of our life. Machines can handle the large amounts
of data and are not tiresome like human beings. The results we receive from machines are
less error prone. So, training a machine to take care of certain health issues is a good idea.
Machine learning understands like how humans will try to understand the data. The only one
basic requirement is to train the machine with right and proper and valid data as those are
machines. If quality data is not provided the machines may get confused and predict the
results with less accuracy. Health issues should be dealt with utmost care. So, the machines
trained with relevant data can predict the diseases in an effective manner.

In order to attain accurate results, the pre-processing of the data needs to be performed
before giving the data to a machine as input.

Pre-processing involves measuring the distance between the data points and finding the
similarity between them using metrics like Euclidean and cosine-based metrics. Mainly
Euclidean distances can be measured using

dEuc (d1, d2) =Sqrt((d1-d2), (d1-d2)) ------------(3)

and Z-score normalization where the data is converted into 0 and 1 where 0 is variance and
standard deviation is 1.

The scaling of features is done in machine learning using two methods (i) Normalization and
(ii) Standardization.

Normalization or min-max scalar is the rescale of the given data in the range of 0 and 1.
Normalization is affected by the presence of outliers. Standardization or z-score is the rescale
of given data objects’ mean to 0 and standard deviation as 1. There are three types of scalars
available into three types like: (i) Standard Scalar (ii) Min-max scalar and (iii) Robust scalar.

2
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

Section 2 of the paper describes the related previous work done already. and Section 3
discusses the proposed method, section 4 provides the results and section 5 concludes the
paper.

2 RELATED WORK
The paper goal is to use blockchain technology to create a decentralised storage system. The
data owner registers themself at first. The owner logs in after successfully registering and
uploads a file using the file picker. The system verifies the file size and guarantees network
storage availability. When there is enough space, the file gets uploaded. The system then
takes action. The AES 256 bit technique is being used to encrypt the submitted file. The
wallet address of the owner and a salt value chosen at random are used to create the
encryption key. The owner's data is encrypted using this encryption key. As a result, the data's
secrecy is preserved. The IPFS protocol is then used to split the encrypted file into 64KB
blocks that are delivered to various peers throughout the network. This gives back a hash
value that contains the file's path information [1].

The system's goal is to experiment with and show a functional secure file storage system for
compact credential files based on blockchain and cryptography. The Kademlia protocol,
which is used in the construction of the P2P network, enables each node to join and leave the
network at any moment, find its nearest neighbours, update its routing table, and search for
any other nodes. In this research, a novel protocol for such a system has been built using
several cryptography and network algorithms [2].

Services for decentralised cloud storage offer a great prospect for a distinct 17 cloud market,
satisfying the supply and demand for IT resources from a large user base. Security concerns
can slow down the realisation of such an opportunity, which is otherwise clearly enticing and
promising for the anticipated economic gains, due to the dynamic and independent nature of
the ensuing infrastructure. Using decentralised cloud services for storage [3].

The user first registers for a metamask account. The program uses web3.js to retrieve the
user's account address and wallet amount from the metamask. Through the file picker, users
choose the file to upload. The system determines how many peers are available. Additionally,
the AES method encrypts the uploaded file using the user's wallet address as a key. A
payment dialogue requests the user's approval. The user's file is saved using the IPFS protocol
across all available peers after the payment is confirmed [4]

The user layer, data processing layer, storage network layer, and blockchain layer are the four
main components of the decentralised storage system proposed in this study. Ordinary users
and administrators are the two categories of users that are registered under the user layer.
Only the administrator agrees to apply for registration as an ordinary user under the system's
access scheme [5].

The distributed cloud storage platform Blockchain data are archived for privacy and security.
Data can be encrypted using the digital hash algorithm on a decentralised cloud with the help
of the private and public keys. Data is kept in peer-to-peer decentralised cloud storage on this
decentralised server. By using a cryptographic technique, it saves encrypted files from both
nodes of the transaction [6].

3
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

A new method for decentralised systems to address flaws such storage constraints and time-
consuming concerns in blockchain and immutably recorded transactions in the IPFS is to
combine IPFS and blockchain. The originality of this study in comparison to others is
communication and interaction outside of blockchain to other decentralised systems. This
study addresses two issues: minimising the issue of slow response times and resolving storage
constraints in blockchain [7-11].

3 METHODOLOGY

System Architecture

Fig. 1. System Design for Decentralized Cloud Storage

In the deployment of a private IPFS network, all nodes and users are interconnected in
a peer-to-peer (p2p) distributed manner. This means that each participant in the network can
communicate directly with others, creating a decentralized and resilient infrastructure. When
a user uploads a file to the IPFS network, the file undergoes a process of fragmentation. The
file is divided into smaller, manageable chunks, typically around 256KB in size by default.
This fragmentation process is essential for efficient distribution and storage across the
network. Before these file chunks are distributed and stored among the nodes, they undergo
an encryption process using the SHA-256 algorithm. SHA-256 is a cryptographic hash
function known for its strong security properties. The encryption ensures that the data
remains confidential and protected from unauthorized access during transit and storage. Once
the file chunks are encrypted, they are distributed to all the peers (nodes/computers) in the
private IPFS network. This distribution process ensures redundancy and fault tolerance, as
multiple copies of each chunk are stored across different nodes. Therefore, even if some
nodes become unavailable or fail, the data remains accessible from other nodes that hold
copies of the same chunks. In the private IPFS network, during the download process, the
user's requested file is retrieved by collecting encrypted file chunks from various distributed
nodes. Once collected, these chunks are decrypted using the appropriate decryption key
generated during the upload. The decrypted chunks are then merged in the correct order to

4
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

reconstruct the original file. Finally, the complete file is delivered to the user in its original
format, ensuring a secure, efficient, and reliable data retrieval experience [9,10].

Central Vs Decentralized

Fig. 2. File transfer in centralized system

Fig. 3. File transfer in decentralized system

Technologies Used :

IPFS ( InterPlanetary File System )

IPFS is a distributed peer-to-peer protocol for decentralized file storage and sharing. It
uses content addressing, assigning unique cryptographic hashes to files based on their
content, enabling efficient retrieval and verification. IPFS employs content caching and
deduplication, storing chunks of files across nodes to enhance availability. It utilizes a
distributed routing system and DHT for efficient file retrieval. IPFS ensures data integrity
through cryptographic verification and supports data versioning. Overall, IPFS provides a
secure, resilient, and decentralized file storage solution with promising applications for
building decentralized systems and applications [11].

Python
Python was the primary programming language used in the project. It facilitated user
interactions, data storage, and communication with the IPFS network. Python’ s object-
oriented nature allowed for efficient database modelling and storage of user information and
file data. It integrated well with the `ipfshttpclient` library, enabling file uploads, retrievals,

5
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

and peer-to-peer connections. Python's robustness ensured error handling and system
stability. Its readable syntax made coding and debugging easier, making Python a suitable
choice for building the secure decentralized cloud platform [8].

Django
Django was the core web framework used in the project. It facilitated user
authentication, data storage, and retrieval through the SQLite database. With Django's
templating engine, dynamic web pages were created, ensuring a smooth user experience. The
framework's URL routing and view functions managed file uploads, downloads, and
interactions with the IPFS network. Overall, Django simplified development and enabled the
implementation of key functionalities in the decentralized cloud platform.

SQLite
SQLite was utilized as the primary database management system in the project. It
served as the backend storage for essential data, including user information, file metadata,
and IPFS hashes. Through the Django ORM (Object-Relational Mapping), the application
interacted seamlessly with the SQLite database. This lightweight and self-contained database
system provided a reliable and efficient solution for managing data in the secure
decentralized cloud platform. Its integration with Django facilitated the handling of user
registration, file uploads, and retrieval, ensuring a robust and scalable foundation for the
application.

Designed System

Swarm Key

In the context of a private network, a swarm key

could be a unique identifier or access token that
grants permission for certain nodes or users to
interact with the Swarm network and access specific
content. It might be used to authenticate nodes before
they can join the private Swarm network or to control
which nodes have read or write access to particular
data. Every node in the private swarm contains
‘swarm.key’ file in the ‘.ipfs’ directory Fig. 4. Swarm.key file

File Upload and Download

The user can easily upload their file through the

user-friendly web interface designed using HTML
and CSS. Once the file is selected and submitted
using the HTML form, Django takes over the
process and seamlessly handles the file. Django
then securely transfers the file to IPFS
(InterPlanetary File System) for decentralized
storage and retrieval. By utilizing this efficient
system, the user can effortlessly manage their files,
ensuring they are securely stored and easily
accessible. This combination of technologies brings
convenience, reliability, and scalability to the file

Fig. 5. Home Page

6
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

handling process, making it a seamless experience for the user.

File Management

The In the secure decentralized cloud

project using IPFS, users can easily
interact with their uploaded files. They
have the ability to view the list of files
they uploaded, download them to their
local machine, and even delete them
from the IPFS network. Additionally,
users can access information about the
peers who are currently storing a
specific file, giving them insights into
the distribution and availability of their
data. These user-friendly features
empower users
to manage their files efficiently and
ensure data privacy and control in the
decentralized cloud environment. User
can view the uploaded files, download
them, or can delete them. They can also
see the peers who are storing that file.

Fig. 6. File Management

File chunk in peer’s computer

The file uploaded by the user is initially split into smaller chunks, and these chunks are
stored in other peers' computers in an encrypted format. Each chunk is further divided into
several small parts, each with a size of 256KB. These smaller parts are then stored in a
folder called 'blocks,' which is located within the '.ipfs' folder. The '.ipfs' folder is present in
the 'Users' directory.

Fig. 7. File chunks in the peer’ system

7
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

Fig. 8. Encrypted File

Analysis

The table above shows the storage costs of a variety of

cloud providers, both centralized and decentralized.
Centralized clouds are owned and operated by a single
entity, while decentralized clouds are distributed across
a network of nodes.
As you can see, decentralized clouds are generally much
more affordable than centralized clouds. Our Cloud, a
new decentralized cloud, offers storage for just
$0.07/TB/month, which is significantly lower than the
cost of any of the centralized clouds in the table. In
addition to being more affordable, decentralized clouds
also offer a number of other advantages, such as greater
security and resilience. Since decentralized clouds are
not controlled by a single entity, they are less vulnerable
to attack or censorship.

Fig. 9. Cloud vs cost table

8
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

Fig. 10. Hypothetical Download speed vs No of peers at constant

The graph depicting download speed vs. the number of peers in a hypothetical IPFS scenario
reveals crucial insights into the behavior of a decentralized peer-to-peer network. Initially, as
the number of peers increases, the download speed experiences a rapid improvement,
highlighting the network's ability to utilize multiple sources for parallel downloads and
reduce latency. However, as the number of peers continues to grow, the rate of improvement
slows down, indicating diminishing returns on download speed. The curve eventually reaches
a saturation point, where further increasing the number of peers doesn't significantly enhance
download speed. This observation emphasizes the need to find an optimal balance between
peer participation and download performance. The graph underscores IPFS's strength in
leveraging its distributed nature for resilience and robustness, ensuring reliable content
retrieval by employing multiple peers. Understanding these patterns is crucial for optimizing
peer networks and improving overall system performance in real-world IPFS deployments.

4 CONCLUSION
In conclusion, our project aimed to explore the concept of decentralized cloud storage using
IPFS (InterPlanetary File System). We developed a web application that allows users to
upload, store, and retrieve files using IPFS technology. Throughout the project, we compared
centralized and decentralized cloud storage models. We identified the advantages of
decentralized storage, such as improved data availability, resilience to failures, and reduced
dependency on a single provider. We also discussed the challenges, including slower network
speeds and higher storage requirements. By implementing IPFS, we leveraged its unique
features like content addressing and distributed peer-to-peer network. This allowed us to
achieve data integrity, censorship resistance, and data redundancy. The files uploaded by
users are split into smaller chunks, encrypted, and distributed across the IPFS network,
ensuring their availability even if some nodes go offline. We integrated additional features
like user registration, authentication, and file management. Users can upload files, view their
uploaded files, download files using IPFS hashes, and delete files when needed. Our project
showcased the potential of decentralized cloud storage and its applicability in real- world
scenarios. It provides an alternative approach to traditional centralized cloud storage, offering

9
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

more control and ownership over data and further opening up the possibility for a sustainable
development in the same aspect. Overall, our project demonstrated the capabilities of IPFS
and highlighted the benefits and challenges associated with decentralized cloud storage. It
opens up avenues for further exploration and research in the field of decentralized
technologies, paving the way for a more decentralized and resilient internet infrastructure.

5 FUTURE SCOPE

The future scope and enhancements of your project include integrating additional
decentralized storage protocols like Filecoin, Sia, or Storj to offer users more storage options.
Improving the user interface and experience with features such as file previews, organization,
search, and error handling can enhance usability. Implementing advanced file management
functionalities like versioning, sharing, and collaborative editing will enable users to track
and collaborate on files effectively. Strengthening security measures with end-to-end
encryption, encryption key management, and multi-factor authentication will enhance data
protection. Optimizing scalability and performance with caching, load balancing, and
distributed storage techniques will ensure efficient handling of user requests. Integrating
blockchain technology can provide added security and transparency. Developing mobile
applications will extend accessibility. Integrating with third-party services and incorporating
community features will enhance collaboration and engagement. Continuous monitoring,
maintenance, and staying updated with advancements in decentralized storage technologies
will ensure a robust, sustainable and reliable system.

6 References
1. G. Richa Shalom, Ganesh Rohit Nirogi, Decentralized cloud storage using Blockchain,
International journal for Research in Applied science & Engineering Technology
(URASET) 2022. ISSN : 2321-9653.
2. Anesh, Sreshta, Bipin, Khatiwada, Sagar, Bhusai, Decentralized secure cloud Storage
using Blockchain, Tribuvan University Institute of Engineering Pulchowk Campus
2022.
3. Mujahid Tabassum, Hanan Aljoaey, Tripati Sharma, Yew Hoe Tung, Secured Data
Storage in Decentralized Cloud by Blockchain Ethereum, Department of Information
Technology, College of Computer, Qassim University. Published on10 November 2022.
ISSN : 2079 9292, DOI - https://doi.org/10.3390/electronics11223686.
4. Shah, Meet, Shaikh, Mohammedhasan, Mishra, Vishwajeet, Decentralized cloud storage
using Blockchain, Department of Information Technology, St. Francis Institute of
Technology, Mumbai, India. Published on June 2020, ISBN:978-1-7281-5519-7, DOI:
10.1109/ICOEI48184.2020.9143004.
5. Lu Meng, Bin Sun, Decentralized Storage Based on a Blockchain, College of
Information Science and Engineering, Northeastern University, Shenyang 110004,
China Published on 12 October 2022, DOI:10.1088/1742-6596/1237/4/042008.
6. Nazmun Nahar, Farah Hasin and Kazi Abu Taher, Application of Blockchain for the
Security of Decentralized Cloud Computing, Department of Information and
Communication Technology Bangladesh University of Professionals Dhaka,
Bangladesh Published on 2021, DOI:10.1109/ICICT4SD50815.2021.9396921.

10
E3S Web of Conferences 430, 01097 (2023) https://doi.org/10.1051/e3sconf/202343001097
ICMPC 2023

7. Morteza Alizadeh, Karl Andersson, and Olov Schelen, Efficient Decentralized Data
Storage Based on Public Blockchain and IPFS”-Department of Computer Science,
Electrical and Space Engineering Lulea University of Technology,
DOI:10.1109/CSDE50874.2020.9411599.
8. Prasanna Lakshmi, K., Reddy, C.R.K. A survey on different trends in Data Streams
(2010) ICNIT 2010 - 2010 International Conference on Networking and Information
Technology, art. no. 5508473, pp. 451-455.
9. Jeevan Nagendra Kumar, Y., Spandana, V., Vaishnavi, V.S., Neha, K., Devi, V.G.R.R.
Supervised machine learning Approach for crop yield prediction in agriculture sector
(2020) Proceedings of the 5th International Conference on Communication and
Electronics Systems, ICCES 2020, art. no. 09137868, pp. 736-741.
10. Sankara Babu, B., Suneetha, A., Charles Babu, G., Jeevan Nagendra Kumar, Y., Karuna,
G. Medical disease prediction using grey wolf optimization and auto encoder based
recurrent neural network (2018) Periodicals of Engineering and Natural Sciences, 6 (1),
pp. 229-240.
11. Nagaraja, A., Boregowda, U., Khatatneh, K., Vangipuram, R., Nuvvusetty, R., Sravan
Kiran, V. Similarity Based Feature Transformation for Network Anomaly Detection
(2020) IEEE Access, 8, art. no. 9006824, pp. 39184-39196

11