ReviewDeep learning and big data technologies for IoT security
Introduction
The swift growth in emerging technologies such as, sensors, smartphones, 5G communication, and virtual reality leads to innovative applications such as, connected industries, smart city, smart energy, connected automobiles, smart agriculture, connected building complexes, connected health care, smart retail outlets, and smart supply chain, which adversely contribute to the accumulation of massive amounts of data. A study conducted by the National Cable & Telecommunications Association (NCTA) predicts that by 2020, approximately 50.1 Billion Internet of Things (IoT) devices will be connected to the Internet. The growth of IoT devices makes the security of these devices debatable [1], [2]
According to McAfee (2018), there has been a barrage of cyberattacks and data breaches that has hit almost every industry since 1st of January 2018. Further, many of these attacks were targeted on IoT devices. The increasing use of IoT devices invites the cybercriminals to target them. Additionally, the prospect of interconnectivity among IoT devices makes them vulnerable [3]. Furthermore, VDC Research Group Inc. have also conducted a study to determine the obstacles in developing connected devices. The research has indicated that 60% of the obstacles are related to security requirements in developing connected devices [4]. In addition, based on Kaspersky Lab’s collection, the number of malware samples for IoT devices has seen a rapid increase from 3219 samples for the year 2016 to 121588 samples for the year 2018. It is clearly evident that there are huge number of vulnerabilities for the IoT devices [5].
According to [2], many organizations are exposed to greatest challenges in monitoring network based threats, prominently in the following sectors: government, energy, healthcare, banks, and research centres. Moreover, these sectors invest in security monitoring tools in order to protect and secure their infrastructure. As mentioned earlier, generally, the IoT devices generate immense amounts of data that flows through networks. Data that flows through a network is at the possible risk for network attacks. Further, the study has argued that the existing tools and techniques are insufficient to detect innovative attacks triggered by cybercriminals due to the volume, velocity, variety, and veracity of data. Moreover, when huge amounts of data are being handled by the network, the security analytics report on a weekly or monthly basis would not be sufficient enough to detect and mitigate the attacks. Furthermore, the study has asserted that big data technologies would be able to handle the challenges of the volume, velocity, variety and veracity of the data.
Data is generally categorized as big data based on the properties associated with it, commonly known as the V’s of big data [6]. Big data technologies are the tools or technologies used to efficiently process these data. Authors of [7], discuss that enterprises collect security related data for regulatory compliance and post hoc forensic analysis. Furthermore, these large enterprises generate approximately 10 to 100 billion events per day. The authors also assert that existing mechanisms lack processing at large scales and big data analytics have been used to analyse and correlate security-related data efficiently and at unprecedented scales.
In this context, this present study proposes to employ deep learning and big data technologies to strengthen the security of IoT devices. Off late, deep learning has gained recognition due to its non-manual feature engineering, unsupervised pre-training, and compression capabilities, these features make the employability of deep learning feasible even in resource constrained networks. Furthermore, deep learning has been widely implemented because of its self-learning capability, potential to yield highly accurate results, and faster processing time. This is vital, as resource constrained system may run into other issues such as out-of-memory access, unsafe programming languages, and so forth [8].
Most of the existing literature separately focuses on deep learning, big data, and IoT security. Some studies have either focused on deep learning [9], [10] or big data [11], [12] for IoT security. To the best of our knowledge, none of the existing studies have comprehensively reviewed the feasibility of employing both of these technologies in context of IoT security.
Table 1 summarizes most of the existing recent relevant studies and highlights the research gap. From Table 1, it is concludable that many studies have failed to consider the impact of volume, velocity, variety, and veracity of data generated by IoT devices, as against [2] who have highlighted the impacts in their study. Hence, inclusion of big data technologies becomes mandatory to address the impact of volume, velocity, variety, and veracity of data generated by IoT devices. Additionally, it is clearly evident in Table 1 that not many studies have focused on deep learning and big data technologies for IoT security.
This paper is intended to guide deep learning, big data, and IoT researchers and developers, to whom IoT security would be of primary concern. The contributions of this paper has been summarized below.
- i
We identified, and highlighted the key issues of IoT security.
- ii
We have picked five IoT security use cases where deep learning and big data technologies could be of potential solution.
- iii
We have surveyed the state-of-the-art researches focused on deep learning, big data technologies, and IoT security, to determine the technical applicability and limitations of these three aforementioned domains.
- iv
We have developed a thematic taxonomy by extracting valuable information from the state-of-the-art.
- v
We have analysed existing solutions based on the derived taxonomy.
- vi
We have highlighted the challenges and have proposed guidelines for future researchers to encourage the successful application of deep learning, big data technologies, and IoT security.
However, this study limits its scope only to deep learning and does not discuss on traditional machine learning algorithms with respect to big data technologies and IoT security. Additionally, this survey also does not go into detail about IoT security for each available smart application area, rather discusses in the networking and communications perspective.
This paper is structured as follows:
Section 2 details the motivation and use cases of, deep learning, big data technologies, and IoT security. Section 3 introduces deep learning, big data technologies, and IoT security. Section 4 provides the thematic taxonomy and discusses its components in detail. Section 5 critically analyses the state-of-the-art studies related to deep learning, big data technologies, and IoT security. Section 6 discusses the challenges and proposes future directions. Finally, Section 7 concludes this present study.
Section snippets
Motivation and use cases
In this section we have detailed on the motivation for our study and provided some use case scenarios that motivate the survey of deep learning and big data technologies for IoT security.
IoT devices have seen rapid growth in recent years, which is of a great concern in terms of the security risks associated with them. The rapid growth of these devices and the availability of modern hacking technologies have forced the necessity to ensure that IoT devices are not vulnerable to security breaches.
Background
This section contains a comprehensive description of deep learning, big data technologies, and IoT security. Additionally, the relationship between these three domains have been discussed, to provide fundamental knowledge and relationship mapping on these leading edge topics.
Taxonomy
This section highlights and proposes a taxonomy for deep learning, big data technologies, and IoT security. This taxonomy is classified into different categories namely, Deep Learning, IoT Security, and Big Data Technologies, and further sub categorized as Deep Learning Architectures, Frameworks, Model Evaluation, IoT Security Application Area, IoT Security Attacks, Datasets, Apache Hadoop, Apache Spark, and Apache Storm. Due to the limited studies that have been conducted by combining deep
State of the art deep learning for IoT security using big data technologies
This section comprises of three subsections. The first subsection presents insights of the state-of-the-art techniques in cases where deep learning has been applied for IoT security. The second subsection details on the application of deep learning along with big data technologies. Finally, a comprehensive review of deep learning, big data technologies and IoT security has been presented.
Open challenges and future directions
This section highlights the most significant research challenges in terms of IoT security using deep learning and big data technologies. The state of-the-art capabilities in IoT security, deep learning, and big data technologies have been examined to determine the major research challenges, suggestions, and future directions.
Conclusion
The expanding population of IoT devices has contributed to the consideration of security risks associated with them. IoT devices are proven to be vulnerable due to the recent increasing attacks such as, the Carna and Mirai botnets. Additionally, IoT devices produce large volume, velocity and variety of data. This makes existing solutions less efficient and require modern day solutions. In this regard, deep learning has been widely accepted amongst researchers and organizations due to their high
Declaration of 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.
Acknowledgements
We would like to extend our sincere gratitude and appreciation to Dr. Ahmed Tajuddin Bin Samsuddin, Mr. Keng Chee Chan, Ir. Dr. Abdul Aziz Bin Abdul Rahman, Mrs. Azlinda Tee Binti Md Azlan Tee, and the members of editorial board, Telekom Research & Development Sdn. Bhd for their continuous support towards the publication of this manuscript. We would also like to thank the anonymous reviewers and the editors of this journal who helped us in improving the quality of our manuscript.
References (182)
- et al.
Enforcing security in internet of things frameworks: A systematic literature review
Internet Things
(2019) - et al.
Cyber security challenges and solutions for v2x communications: A survey
Comput. Netw.
(2019) - et al.
Smart grids security challenges: Classification by sources of threats
J. Electr. Syst. Inf. Technol.
(2018) - et al.
Distributed attack detection scheme using deep learning approach for internet of things
Future Gener. Comput. Syst.
(2018) - et al.
A framework for fast and efficient cyber security network intrusion detection using apache spark
Procedia Comput. Sci.
(2016) - et al.
Deep learning for visual understanding: A review
Neurocomputing
(2016) - et al.
Deep learning for sensor-based activity recognition: A survey
Pattern Recognit. Lett.
(2019) - et al.
Beyond the hype: Big data concepts, methods, and analytics
Int. J. Inf. Manag.
(2015) - et al.
Securing iots in distributed blockchain: Analysis, requirements and open issues
Future Gener. Comput. Syst.
(2019) - et al.
Iot security: Review, blockchain solutions, and open challenges
Future Gener. Comput. Syst.
(2018)
Securing against brute-force attack: A hash-based rfid mutual authentication protocol using a secret value
Comput. Commun.
Perception layer security in internet of things
Future Gener. Comput. Syst.
Modeling word perception using the elman network
Neurocomputing
Autoencoder for words
Neurocomputing
Breast cancer classification using deep belief networks
Expert Syst. Appl.
Deep learning and software-defined networks: Towards secure iot architecture
Internet Things
A deep recurrent neural network based approach for internet of things malware threat hunting
Future Gener. Comput. Syst.
DRTHIS: Deep ransomware threat hunting and intelligence system at the fog layer
Future Gener. Comput. Syst.
Edge-fog cloud: A distributed cloud for internet of things computations
Real-time big data processing for anomaly detection: A survey
Int. J. Inf. Manage.
Trending: Iot malware attacks of 2018
Developers discuss iot security and platforms trends
New trends in the world of iot threats
Big data: Issues, challenges, tools and good practices
Big data analytics for security
IEEE Secur. Priv.
Botnet detection in the internet of things using deep learning approaches
Cybersecurity challenges and opportunities in the new edge computing+ iot world
Securing sensor to cloud ecosystem using internet of things (iot) security framework
Big data privacy in the internet of things era
IT Prof.
Review of different encryptionand decryption techniques used for security and privacy of iot in different applications
Deep learning for iot big data and streaming analytics: A survey
IEEE Commun. Surv. Tutor.
Big iot data analytics: architecture, opportunities, and open research challenges
IEEE Access
Securing the internet of things: Challenges, threats and solutions
Internet Things
Security attacks in iot: A survey
Deep learning for classification of malware system call sequences
Droid-sec: deep learning in android malware detection
Droiddetector: android malware characterization and detection using deep learning
Tsinghua Sci. Technol.
N-baiot—network-based detection of iot botnet attacks using deep autoencoders
IEEE Pervasive Comput.
A deep learning approach for intrusion detection in internet of things using bi-directional long short-term memory recurrent neural network
B-dids: Mining anomalies in a big-distributed intrusion detection system
A big data framework for intrusion detection in smart grids using apache spark
Deep learning approach for intelligent intrusion detection system
IEEE Access
Sirenjack attack lets hackers take control over emergency alert sirens
5 biggest iot security failures of 2018
Hackers use ddos attack to cut heat to apartments
Iot role in dyn cyberattack
Large ddos attacks cause outages at twitter, spotify, and other sites – techcrunch
The possible vendetta behind the east coast web slowdown
How a fish tank helped hack a casino
Cited by (229)
Deep learning-enabled polar code decoders for 5G networks and beyond
2024, AEU - International Journal of Electronics and CommunicationsCSformer: Enhancing deep learning efficiency for intelligent IoT
2024, Computer CommunicationsA new concatenated Multigraph Neural Network for IoT intrusion detection
2023, Internet of Things (Netherlands)A survey of UAV-based data collection: Challenges, solutions and future perspectives
2023, Journal of Network and Computer Applications