PervasiveFL: Pervasive Federated Learning for Heterogeneous IoT Systems

In this technology era, the Internet of Things (IoT) plays predominant role in several aspects of real-life problem. The thrust area of IoT applications include smart city, smart transport, smart healthcare, smart environment, smart agriculture, etc. The IoT system involves heterogeneous sensors and devices at the edge layer of the protocol stack. These IoT devices are characterized in term of hardware specification, different communication protocols, services offered by these devices and trustable users of the IoT devices. Hence, the IoT requires efficient security and privacy mechanism for heterogeneous IoT devices and as well as voluminous amount of data generated by these IoT devices. Conventionally, the security for homogenous devices is provided by the centralized mechanism through authentication, encryption/decryption of data, digital signatures and cryptographic algorithms. However, IoT has characteristics of heterogeneous, autonomous and resource constraint IoT devices, in which case, the conventional centralized security solutions fail tremendously. Hence, the decentralized approach of Block chain is well suited for this scenario. This chapter targets to provide the comprehensive review of various security, privacy and access issues, solution approaches and challenges towards IoT. The role block chains in five different IoT applications are discussed in this chapter. These IoT applications include Intelligence Transport system, Smart Healthcare system, supply chain management, IoT ecosystem, Smart City. This chapter discusses how block chain provides security, smart contracts, access control and proof-of-work and proof-of-state are carried out at decentralized manner. The chapter presents the use case of block chain beyond crypto currency and bit coin, the data are to be preserved through block chain technique and different mining techniques used in block chain for these IoT applications.

The proliferation of the Internet of Things (IoT) has since seen a growing interest in architectural design and adaptive frameworks to promote the connection between heterogeneous IoT devices and IoT systems. The most widely favoured software architecture in IoT is the Service Oriented Architecture (SOA), which aims to provide a loosely coupled systems to leverage the use and reuse of IoT services at the middle-ware layer, to minimise system integration problems. However, despite the flexibility offered by SOA, the challenges of integrating, scaling and ensuring resilience in IoT systems persist. One of the key causes of poor integration in IoT systems is the lack of an intelligent, connection-aware framework to support interaction in IoT systems. This paper reviews existing architectural frameworks for integrating IoT devices and identifies the key areas that require further research improvements. The paper concludes by proposing a possible solution based on microservice. The proposed IoT integration framework benefits from an intelligent API layer that employs an external service assembler, service auditor, service monitor and service router component to coordinate service publishing, subscription, decoupling and service combination within the architecture.

The evolution of the Internet of things (IoT) has started from simple machines that display temperature and goes all the way to powerful and sophisticated machines that can take care of individual buildings. This has made IoT devices one of the most desirable targets as botnets. Infecting IoT devices and controlling them as botnets happens quite often. This is mostly due to the lack of proper security standards and defense mechanisms in IoT devices. Common ways that attackers convert IoT devices into botnets include using default or no login passwords, keeping old firmware with no updates or security patches, and using same vulnerable software by multiple manufacturers. IoT botnets, therefore, are being used for Distributed Denial of Services (DDOS) attacks on a massive scale, reaching up to Terabytes of data per second. The rise of IoT and IoT-based botnets has therefore led to new security and management issues. To address these issues, this chapter proposes a microservice architecture-based solution dealing with the problem of heterogeneous IoT platforms and the lack of IoT device monitoring for security and fault tolerance. This chapter first describes the different types of IoT systems and explains the main reasons behind the progress of IoT botnets. It then describes potential IoT botnet architectures, and how botnets may be detected. It illustrates a data pipeline for IoT devices to send data through a messaging platform Kafka and monitor the devices using the collected data by making real-time dashboards. For proof of concept, the proposed solution is tested on a heterogeneous IoT cluster, including Raspberry Pi’s and IoT devices from different vendors. The heterogeneous IoT system is turned into botnets by a controlled variation of Mirai botnet malware and by conventional Metasploit hack on the Raspberry Pi device. Some simple machine learning method is then deployed for data analysis to detect if the IoT device is infected and turned into a botnet. Simple experimental results validate the design. The chapter ends with a conclusion including suggestions for future directions on detecting and preventing new IoT-based botnets.

Distributed training of Machine Learning models in edge Internet of Things (IoT) environments is challenging because of three main points. First, resource-constrained devices have large training times and limited energy budget. Second, resource heterogeneity of IoT devices slows down the training of the global model due to the presence of slower devices (stragglers). Finally, varying operational conditions, such as network bandwidth, and computing resources, significantly affect training time and energy consumption. Recent studies have proposed Split Learning (SL) for distributed model training with limited resources but its efficient implementation on the resource-constrained and decentralized heterogeneous IoT devices remains minimally explored. We propose Adaptive REsource-aware Split-learning (ARES), a scheme for efficient model training in IoT systems. ARES accelerates training in resource-constrained devices and minimizes the effect of stragglers on the training through device-targeted split points while accounting for time-varying network throughput and computing resources. ARES takes into account application constraints to mitigate training optimization tradeoffs in terms of energy consumption and training time. We evaluate ARES prototype on a real testbed comprising heterogeneous IoT devices running a widely-adopted deep neural network and dataset. Results show that ARES accelerates model training on IoT devices by up to 48% and minimizes the energy consumption by up to 61.4% compared to Federated Learning (FL) and classic SL, without sacrificing model convergence and accuracy.

The Internet of Things (IoT) technology has revolutionized various industries by allowing data collection, analysis, and decision-making in real time through interconnected devices. However, challenges arise in implementing Federated Learning (FL) in heterogeneous industrial IoT environments, such as maintaining model accuracy with non-Independent and Identically Distributed (non-IID) datasets and straggler IoT devices, ensuring computation and communication efficiency, and addressing weight aggregation issues. In this study, we propose an Uncertainty-Aware Federated Reinforcement Learning (UA-FedRL) method that dynamically selects epochs of individual clients to effectively manage heterogeneous industrial IoT devices and improve accuracy, computation, and communication efficiency. Additionally, we introduce the Predictive Weighted Average Aggregation (PWA) method to tackle weight aggregation issues in heterogeneous industrial IoT scenarios by adjusting the weights of individual models based on their quality. The UA-FedRL addresses the inherent complexities and challenges of implementing FL in heterogeneous industrial IoT environments. Extensive simulations in complex IoT environments demonstrate the superior performance of UA-FedRL on both MNIST and CIFAR-10 datasets compared to other existing approaches in terms of accuracy, communication efficiency, and computation efficiency. The UA-FedRL algorithm attain an accuracy of 96.83% on the MNIST dataset and 62.75% on the CIFAR-10 dataset, despite the presence of 90% straggler IoT devices, attesting to its robust performance and adaptability in different datasets.

One of the important characteristics envisioned for 6G is security function virtualization (SFV). Similar to network function virtualization (NFV) in 5G networks, SFV provides new opportunities for improving security while reducing the security overhead. In particular, it provides an attractive way of solving compatibility issues related to security. Malware in Internet of Things (IoT) systems is gaining popularity among cyber-criminals because of the expected number of IoT devices in 5G and 6G networks. To solve this issue, this article proposes a security framework that exploits softwarization of security functions via SFV to improve trust in IoT systems and contain the propagation of malware. IoT devices are categorized into trusted, vulnerable, and compromised levels using remote attestation. To isolate the devices in the three distinct categories, NFV is used to create separate networks for each category, and a distributed ledger is used to store the state of each device. Virtualized remote attestation routines are employed to avoid any compatibility issues among heterogeneous IoT devices and effectively contain malware propagation. The results show that the proposed framework can reduce the number of infected devices by 66 percent in only 10 seconds.

Efficient hybrid centralized and blockchain-based authentication architecture for heterogeneous IoT systems

The rapid development of Internet of Things (IoT) systems has led to the problem of managing and analyzing the large volumes of data that they generate. Traditional approaches that involve collection of data from IoT devices into one centralized repository for further analysis are not always applicable due to the large amount of collected data, the use of communication channels with limited bandwidth, security and privacy requirements, etc. Federated learning (FL) is an emerging approach that allows one to analyze data directly on data sources and to federate the results of each analysis to yield a result as traditional centralized data processing. FL is being actively developed, and currently, there are several open-source frameworks that implement it. This article presents a comparative review and analysis of the existing open-source FL frameworks, including their applicability in IoT systems. The authors evaluated the following features of the frameworks: ease of use and deployment, development, analysis capabilities, accuracy, and performance. Three different data sets were used in the experiments—two signal data sets of different volumes and one image data set. To model low-power IoT devices, computing nodes with small resources were defined in the testbed. The research results revealed FL frameworks that could be applied in the IoT systems now, but with certain restrictions on their use.

In recent years, the Internet of Things (IoT) is growing rapidly and is being applied in a variety of industries including healthcare, smart homes, and smart cities. Many standard IoT platforms are proposed to connect and communicate with IoT devices easily and securely such as oneM2M, GoogleWeave and Apple HomeKit. However, this makes IoT application development difficult as it requires IoT devices and applications to support multiple protocols to connect different IoT platforms. Therefore, it is necessary to provide a consistent schema to support interoperability in heterogeneous IoT networks. In this paper, we propose how to design and implement interoperating schema between two edge servers oneM2M and Open Connectivity Foundation (OCF) with heterogeneous IoT devices. Specifically, we build proxies for bridging oneM2M Mobius edge server and OCF IoTivity edge server. The sensor data is collected from various IoT devices and sent to an edge server with a compatible platform. Next, the data stored in each server will be exchanged with the other server through a proposed interworking proxy. We also use a rules engine to automatically identify registered devices to support edge server interaction within the same domain and across domains. In addition, we build a web application in each edge server to provide friendly IoT services (data visualization) to clients from different environments. In order to evaluate our system, we collect the delay time of each process in the edge servers. The results show that our proposal is completely applicable in practice.

How to share and exchange information between heterogeneous IoT (Internet of Things) systems is a big problem in building smart cities. Based on general demand for information Sharing & Exchange between IoT systems, this paper proposed a novel system architecture and designed two kinds of deployment model for information Sharing & Exchange and workflows. It could provide unified guideline for building information Sharing & Exchange system between IoT application systems in smart cities.

Internet-of-Things (IoT) applications have been rapidly deployed into pervasive environment, where both challenges and opportunities abound. On the one hand, a large number of IoT devices and their rich functions contribute significant volumes of data, which has brought tremendous convenience to the daily lives of end users. On the other hand, the heterogeneous IoT devices and a large amount of private information transmitted through networks also bring serious security and privacy issues. It is a big challenge to model IoT systems and trust relationships between different entities with a large number of heterogeneous IoT devices. In this article, we study a general IoT system architecture with consideration of heterogeneous IoT devices. Different trust models are proposed and analyzed based on the trust relationships between different entities in the IoT system. We propose a flexible and efficient authentication scheme with a consideration of heterogeneous IoT devices based on the least trust-required model. The proposed scheme provides security and privacy to resource-limited IoT devices flexibly and efficiently by utilizing IoT devices with better storage and computational ability. Moreover, secure data transmission is presented with contextual privacy and data integrity services. The proposed scheme achieves not only the mutual authentication, initial session key agreement, and data integrity but also anonymity, contextual privacy, forward security, end-to-end security, and key escrow resilience. Security analysis is presented to provide verification of the proposed protocol and security objectives. Moreover, performance evaluation is presented with comparison to the other schemes in terms of security features, computational overhead, and communication overhead. The performance comparisons show that our proposed scheme provides flexible and efficient security by consideration of heterogeneous IoT devices. With the higher proportion of resource-limited IoT devices, our proposed scheme outperforms other similar schemes.

Federated learning (FL) is the up-to-date approach for privacy constraints Internet of Things (IoT) applications in next-generation mobile network (NGMN), 5 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> generation (5G), and 6 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> generation (6G), respectively. Due to 5G/6G is based on new radio (NR) technology, the multiple-input and multiple-output (MIMO) of radio services for heterogeneous IoT devices have been performed. The autonomous resource allocation and the intelligent quality of service class identity (IQCI) in mobile networks based on FL systems are obligated to meet the requirements of privacy constraints of IoT applications. In massive FL communications, the heterogeneous local devices propagate their local models and parameters over 5G/6G networks to the aggregation servers in edge cloud areas. Therefore, the assurance of network reliability is compulsory to facilitate end-to-end (E2E) reliability of FL communications and provide the satisfaction of model decisions. This paper proposed an intelligent lightweight scheme based on the reference software-defined networking (SDN) architecture to handle the massive FL communications between clients and aggregators to meet the mentioned perspectives. The handling method adjusts the model parameters and batches size of the individual client to reflect the apparent network conditions classified by the k-nearest neighbor (KNN) algorithm. The proposed system showed notable experimented metrics, including the E2E FL communication latency, throughput, system reliability, and model accuracy.

Several challenges arises in any Smart City ecosystem, such as the governance (management) of heterogeneous IoT (Internet of Things) systems deployed and managed by different entities. Usually, IoT-based smart city applications operate in isolation, leading to a fragmented IoT landscape that hinders the managements of the IoT systems in an integrated way. Thus, it is required the involvement of city authorities to structure and organize city-wide interventions among the different IoT systems/applications; but also it is needed the infrastructure and technology that allow these city-wide interventions between the different levels of government (from upper tier authorities to down tier) for the proper management of heterogeneous IoT systems deployed within a Smart City by public as well as private entities. In this paper we propose an infrastructure for the Smart Governance of heterogeneous IoT solutions within Smart Cities, which provides useful features such as the spread of power, security constraints and scalability, among others. The infrastructure is described through 1) a new architectural model, as an adapted version of the basic architectural model, 2) the use of Digital Objects as crucial artifacts to represent, store and interact with physical and digital elements and 3) the use of open-source technology to implements the architectural model and supports the management of the Digital Objects.

The emergence of various smart services delivered by heterogeneous Internet of Things (IoT) devices has made daily human-life easy and comfortable. IoT devices have brought enormous convenience to various applications, no matter the IoT systems include homogeneous devices like in most sensor networks or heterogeneous devices like in smart homes or smart business applications. However, several known communication infrastructures of IoT systems are at risk to various security attacks and threats. The practice of discovering uncommon occurrences of conventional behaviors is known as anomaly detection. It is an essential tool for detecting fraud as well as network intrusion. In this work, we provide an anomaly-based model on the Extended Isolation Forest method. In our work, the available dataset ’UNSW_2018_IoT_Botnet_Final_10_best_Testing’ has been used for the experiment. Performance indicators, including accuracy, precision, recall, and F1-Score, are used to validate the performance of our suggested system. We get an Accuracy Score of 93% and F1-Score of 96% through the experiment. In addition, the most important top 12 features have a more substantial impact on correct prediction for anomaly identification and have also been identified in this study.KeywordsAnomaly detectionIsolation forestExtended isolation forestIot securityFeature set

As the Internet of Things (IoT) continues to expand, the protection of user data privacy and the security of IoT systems have become increasingly critical. This paper proposes a secure framework that integrates Federated Learning (FL) with advanced cryptographic techniques to address privacy concerns while enabling collaborative machine learning across heterogeneous IoT devices. The proposed framework allows local data processing on IoT devices, ensuring that sensitive information remains decentralized and is never exposed to potential breaches during model training. By employing techniques such as Additive Homomorphic Encryption (AHE) and Secure Multi-Party Computation (SMPC), our approach enables the secure aggregation of model updates while maintaining data confidentiality and integrity. Furthermore, the framework effectively minimizes communication overhead and computational demands, making it suitable for resource-constrained IoT environments. This study not only enhances privacy protections and data security but also improves the efficiency of data analytics in smart cities, healthcare, and industrial applications. Future work will investigate the scalability and adaptability of this framework in real-world scenarios, paving the way for a more secure and privacy-oriented IoT ecosystem.