Ebook: Advances in Edge Computing: Massive Parallel Processing and Applications

With the rapid advances in Internet of Things (IoT) technologies, the number of IoT connected devices at the Edges of the Internet is exponentially growing, accounting for billions of connected devices world-wide. While this is bringing unforeseen opportunities to all fields of science, engineering, businesses and everyday life, many challenges have also risen and need to be addressed by researchers, developers and engineers in the field. Among such challenges, we can distinguish the architectural challenge and the data deluge challenge. As for the former, the exponential growth of computing devices at the Edges of the Internet has created an architectural bottleneck due to a very large number of IoT devices connected to a rather small number of servers in Cloud data centers. The latter is due to data generation by IoT devices, which constitute a main source for data sensing and data streaming leading to Big Data and Big Data streams. Other challenges at the Edges of Internet include leveraging small data centers and data storage capabilities, security, privacy, etc.

Edge computing aims to alleviate the computational burden of the IoT for the Cloud by pushing part of the computations and logics of processing from the Cloud to the Edges of the Internet. At this layer, it is now commonplace to allocate tasks and applications such as data filtering, classification, semantic enrichment and data aggregation, among others. In order to support the pre-processing of data generated from the IoT layer at high rates, the Edge computing layer needs to efficiently implement massively parallel and distributed algorithms, otherwise this new layer would become another bottleneck for the whole computing stack from IoT to the Cloud. While already known parallel, scheduling and allocation algorithms are also implementable for Edge computing, the limited computing capacity as well as the variety and heterogeneity of computing resources at the Edges of Internet require deeper understanding and innovative solutions. Likewise, issues arising from integration and communication of this new wide layer with IoT and the Cloud at large scale need further investigation.

Researchers and developers are currently seeking to achieve a twofold view of Edge Computing, namely, that of computing fabric and data fabric. Altogether, the objective is to create pools of computing, network, communications and storage resources configurable into a scalable fabric that can process, analyse and store massive amounts of data, both offline and online. Indeed, the view of computing fabric is suitable for the Edges of Internet to maximize the usage of Edge computing resources, close to the end- users, for accommodating allocation of tasks of varying sizes and complexity. Therefore, IoT systems can benefit from this computing fabric flexibility whereby applications can be allocated to the suitable pool of resources they need to for their execution and completion. Likewise, the data fabric view can support IoT data stream processing and applications at the Edges of Internet by faster access to data, Edge cache for caching of Internet-based content in support to real time applications of virtual reality, critical decision making, etc., requiring low latency.

The vision of computing and data fabric of Edges of Internet is supported by 5G technologies, Mobile Edge Computing (MEC) servers, micro-data centers, which can improve substantially Quality of Service (QoS), reduce significantly latency (when compared to Cloud latency), provide support for mobility, scalability, real time applications, etc.

This book brings to readers a number of contributions on Edge Computing from a multi-disciplinary perspective. Fundamental research as well as applications from various domains such as smart homes, eHealth, robotics and smart cities are presented. Issues and challenges in respective domains are also identified and discussed towards envisioning efficient and scalable solutions at the Edges of Internet.

The book comprises 11 chapters, which are arranged and summarized as follows.

The first chapter “The Vision of Edges of Internet as a Computing Fabric” by Xhafa discusses the main concepts behind Edge computing as a compute fabric to support real-time applications, high performance computing, big data and big data stream processing at large scale. Besides main concepts, IoT data stream processing, semantic data enrichment, task and application allocation, semantic resource categorization and clustering at the Edges of Internet and introduced and analyzed. The challenges of achieving compute and data fabric vision due to a number of complexities are also discussed.

Pisani et al. in the second chapter “Fog Computing on Constrained Devices: Paving the Way for the Future IoT” present a survey of the concepts of constrained devices, IoT, and fog and mist computing. A classification of applications according to the amount of resources they require (e.g., processing power and memory) is also provided. The authors also discuss what can be expected in a future, where constrained devices and fog computing are used to push the IoT to new limits as well as some challenges and opportunities that these technologies may bring.

The third chapter by Confais et al. “A Fog storage software architecture for the Internet of Things” motivate the need for an operational Fog architecture that supports low latency, mobility (of users and possibly infrastructure) and network partitioning. The authors discuss and analyze in detail such an architecture that consists in coupling an object storage system and a scale-out Network Attached Storage allowing both scalability and performance. Additionally, a new protocol inspired from the Domain Name System to manage replicas in a context of mobility is presented. Finally, the authors discuss the conceptual proximity between Fog storage, Content Delivery Networks and Name Data Networking. A complete experimental evaluation is done over the French cluster Grid’5000 to support the findings with empirical results.

In the fourth chapter, “Towards an Architecture for Information-Centric Network with Content Replication in Edge Computing”, Nascimiento et al., present and analyze the advantages of Information-Centric Networks (ICN) as a new internet architecture and a content distribution model for the Edges of Internet. A discussion towards an architecture using cache replication for ICN networks through Software-Defined Networking (SDN) in Edge computing future scenarios is also included.

Singh et al. in the fifth chapter, “A Quality-Assuring, Combinatorial Auction Based Mechanism for IoT-Based Crowdsourcing” address some research issues from IoT-based crowdsourcing in a strategic setting. The authors have considered the scenario in the IoT-based crowdsourcing, where there are multiple task requesters and multiple IoT devices as task executors with tasks having start and finish times. The objective is to allocate the subset of IoT devices to the tasks, distributed into different slots, in a non-conflicting manner while maximizing the social welfare. The concept of peer grading and the truthful mechanism design are used to that end.

The sixth chapter “Edge Intelligence and the Industrial Internet of Things”, the authors Hill and Al-Aqrabi examine the Edge intelligence in the context of the Industrial Internet of Things as a key enabler of a new set of capabilities in the manufacturing industry. Edge intelligence is seen as a consequence of the desire to “push” computation and storage to the limits of network infrastructure, in a way that supports intelligent visualization and interaction with data within the context that it is generated. Pertinent topics are analysed for potential consumers of the Industrial Internet of Things (IIoT), and by way of a case study in the manufacturing domain, the limits and potential of edge intelligence are explored.

Tinamas and Natwichai in the seventh chapter “Issues in Privacy Preservation for Republishable Data” are concerned with privacy issues related to re-publishing and changing of data in a volatile environment such as edge computing model with distributed data sources. The authors start by presenting the background of privacy preservation environment and then formulate he re-publishable dataset problem. An analysis of possible privacy breach is illustrated. Anonymization techniques including k-anonymity and l-diversity and m-distinct are analysed and discussed.

In the eighth chapter “A study on various cryptographic techniques towards fog environment”, Mahendran et al., various existing security mechanisms in fog computing such as Ciphertext-Policy Attribute-Based Encryption (CP-ABE), Black-Box Traceable CP-ABE, Lightweight Fine-Grained ciphertext search as well as the security mechanism based on evolutionary game theory, the hybrid scheme for fine-grained search and access authorization, variant of password-authenticated key exchange protocol, fog computing through the blockchain and quantum-based steganography protocol for fog are discussed and compared. Likewise, techniques for trust and authentication-based attacks with the future research challenges are analysed. The chapter concludes with a discussion on the advantages and limitations of other security techniques and a summary of the existing research challenges in fog security.

The ninth chapter “iHome: Secure Smart Home Management in the Cloud and the Fog” by Petrakis and Myrizakis present iHome, a smart home management service that runs in the Cloud, implemented based on the principles of Service-Oriented Architecture (SOA) design as a composition of RESTful services. The service addresses the need of users to monitor and control their homes remotely provided that the home devices are “smart” themselves. To mitigate concerns of data protection, response time and communication delays in delivering large amounts of data to the cloud, all services for the home are realized within a fog node installed at home. Data publication and subscription services, an innovative feature of iHome, is a rule-based event management service which forwards alerts to subscribed users for responding to critical events such as incidents of fire, malfunctioning appliances at home, etc.

Goli at al. in the tenth chapter “Hybrid Neural Network and Improved Cuckoo Optimization Algorithm for Forecasting Thermal Comfort Index at Urban Open Spaces”, study the thermal comfort of urban open spaces as a relevant and complex topic in the field of edge and fog computing. The prediction of thermal comfort is significant in order to enable planning of the usage time of urban open spaces. The chapter develops an Improved Cuckoo Search algorithm for forecasting physiological equivalent temperature (PET) values one hour ahead. A proper strategy for tuning the parameters is presented. Moreover, the generation of laid eggs is done by implementing the cross-over operator of a Genetic Algorithm (GA). Then, it is employed to train feed forward neural networks for PET prediction. Finally, the performance of the proposed algorithm is compared to the state-of-the-art; i.e., traditional Cuckoo Optimization Algorithm (COA) and Genetic Algorithm (GA). The reported simulation results demonstrate the effectiveness of the proposed algorithm.

The last chapter “A Cloud Robotic Solution to Assist Medical Application” by Harinee Shan and Anand Mahendran deal with performance and cost issues of conventional robots used in the medical field needing high memory and computation power for decision making and task execution. The authors propose a new architecture which makes use of Commercial Off-The-Shelf (COTS) for cloud robotics for assisting patients who need first aid at times of emergency (based on the data from COTS product). COTS help in tracking the health of individuals and these reports are stored either in the devices or on storage spaces. The authors discuss how to address the memory requirement for the robot as well as the cost reduction on the memory of the robot and of its size.

The Book Editors

Prof. Dr. Fatos Xhafa

Visiting Professor

Department of Computer Science, University of Surrey, UK

On Leave from Universitat Politècnica de Catalunya, Barcelona, Spain

Dr. Arun Kumar Sangaiah

Vellore Institute of Technology, Vellore, India

Edge computing has emerged as a major disrupting technology after Cloud computing to fill in the computational and infrastructural gaps in IoT and Mobile Cloud computing. Indeed, IoT and Mobile Cloud computing are based on rather direct connection of devices to Cloud servers and data centers, such as through gateways, and are not able, due to round trip time, to cope with demanding requirements of real-time applications for low latency, critical decision making, increased security, support to mobility, etc. The aim of computing at Edges of Internet is to alleviate the burden of IoT data stream processing to the Cloud computing by pushing part of the computations, storage, reasoning and intelligence to the Edges of the Internet, close to where data is generated and to end-users. In this introductory chapter we discuss the vision of Edges of Internet as a computing fabric to support real-time applications, high performance computing, big data and big data stream processing at large scale. The ever growing number of compute nodes (from small to large), of fast connectivity (supported by 5G technologies) and of data storage (mini/nano data centers) provide the basis to achieve the vision of computing fabric. The challenges of achieving such vision due to a number of complexities are also discussed.

In the long term, the Internet of Things (IoT) is expected to become an integral part of people’s daily lives. In light of this technological advancement, an ever-growing number of objects with limited hardware may become connected to the Internet. In this chapter, we explore the importance of these constrained devices as well as how we can use them in conjunction with fog computing to change the future of the IoT. First, we present an overview of the concepts of constrained devices, IoT, and fog and mist computing, and then we present a classification of applications according to the amount of resources they require (e.g., processing power and memory). After that, we tie in these topics with a discussion of what can be expected in a future where constrained devices and fog computing are used to push the IoT to new limits. Lastly, we discuss some challenges and opportunities that these technologies may bring.

The last prevision of the european Think Tank IDATE Digiworld estimates to 35 billion of connected devices in 2030 over the world just for the consumer market. This deep wave will be accompanied by a deluge of data, applications and services. Thus, it is quite urgent to propose operational Fog architectures that support low latency, mobility (of users and possibly infrastructure) and network partitioning. In this chapter, we will detail such an architecture that consists in coupling an object store system and a scale-out NAS (Network Attached Storage) allowing both scalability and performance. Moreover, we provide a new protocol inspired from the Domain Name System (DNS) to manage replicas in a context of mobility. Finally, we discuss the conceptual proximity between Fog storage, Content Delivery Networks (CDN) and Name Data Networking (NDN). A complete experimental evaluation is done over the French cluster Grid’5000 in a second part of the chapter.

The traffic reduction in network segments through content network implementations has become a major research topic due to the exponential increase in data requests through the Internet. Even with high-speed connections TCP/IP model still depends on end-to-end communication between two systems, making requests and responses operations expensively for the data link. Therefore Information-Centric Networks (ICN) is a new internet architecture that has received considerable attention. This model has been widely discussed as a new content distribution model for the Internet. To provide improved network management many approaches are using Software-Defined Networking (SDN) to develop flexible content-based networks. This work proposes a discussion towards an architecture using cache replication for ICN networks through SDN in edge computing.

In this chapter, we study some research issues from IoT-based crowdsourcing in a strategic setting. We have considered the scenario in IoT-based crowdsourcing, where there are multiple task requesters and multiple IoT devices as task executors. Each task requester has multiple tasks, with the tasks having start and finish times. Based on the start and finish times, the tasks are to be distributed into different slots. On the other hand, in each slot, each IoT device requests for the set of tasks that it wants to execute along with the valuation that it will charge in exchange for its service. Both the requested set of tasks and the valuations are private informations. Given such scenario, the objective is to allocate the subset of IoT devices to the tasks in a non-conflicting manner with the objective of maximizing the social welfare. For the purpose of determining the unknown quality of the IoT devices we have utilized the concept of peer grading. Therefore, we have designed a truthful mechanism for the problem under investigation that also allows us to have the true information about the quality of the IoT devices.

The Industrial Internet of Things is a key enabler of a new set of capabilities for the manufacturing industry. As more objects become inter-connected, there is a greater need to a) securely share data and b) intelligently manage and exploit the potential of data that is accessible close to its source. Edge intelligence is a consequence of the desire to “push” computation and storage to the limits of network infrastructure, in a way that supports intelligent visualisation and interaction with data within the context that it is generated. This chapter examines pertinent topics for potential consumers of IIoT, and by way of a case study in the manufacturing domain, explores the limits and potential of edge intelligence.

In order to preserve the privacy in the scenarios where the data can be changed by the insert, update, and delete operations, at all times or re-publishable situation, the non-static existing approaches and algorithms may not be appropriate. The changing of data in a volatile environment, e.g. edge computing model with distributed datasources, could even escalate the issue in term of effectiveness and efficiency. In this paper, we elaborate the issues to address such problem. First, the background of privacy preservation is presented. Subsequently, we show the environment of the re-publishable dataset problem. Then, the analysis of possible privacy breach is illustrated. Finally, the paper is summarized and the problem is formally presented and discussed.

Fog is a computing environment which extends the resources to the end of the network infrastructure. It reduces the latency and reachability to the user. Fog computing supports heterogeneity, mobility, and wireless access. Fog nodes had distributed over the vast geographical area. These features facilitate fog computing for sensitive real-time applications which concerns about low latency, such as smart city, smart home, healthcare, and other industrial applications. Simultaneously, these features of fog computing create a space for virtualization issues, web security issues, internal/external communication issues, data security-related issues, wireless security issues, and malware protection. In this chapter, the various existing security mechanisms in fog computing such as Ciphertext-Policy Attribute-Based Encryption(CP-ABE), Black-Box Traceable CP-ABE(T-CP-ABE), Lightweight Fine-Grained ciphertext search(LFGS), security mechanism based on evolutionary game theory, hybrid scheme for fine-grained search and access authorization, variant of password-authenticated key exchange (vPAKE) protocol, fog computing through blockchain and quantum based steganography protocol for fog are discussed and compared. Next, the proposed techniques for trust and authentication based attacks with the future research challenges are tabulated. Then, the advantages and limitations of the other techniques are discussed. Finally, the existing research challenges in fog security are summarized.

iHome is a smart home management service that runs in the cloud. The service addresses the need of users to monitor and control their homes remotely provided that the home devices are “smart” themselves (i.e. they can be connected to the internet and operated remotely). Home devices transmit their identifier, measurements and status to a fog node and from there to the cloud. To mitigate concerns in regards to data protection, response time and communication delays in delivering large amounts of data to the cloud, all services for the home are realized within a fog node installed at home. This information becomes available to registered users in the cloud based on subscriptions (i.e. to users authorized to review and respond to this information). User access rights are defined based on user roles (i.e. cloud administrators, home moderators and residents). Besides data publication and subscription services, an innovative feature of iHome, is a rule-based event management service which forwards alerts to subscribed users for responding to critical events (i.e. incidents of fire, malfunctioning appliances at home). iHome is implemented based on principles of Service Oriented Architecture design as a composition of RESTful services.

Edge and fog computing mainly deal with Internet of Things (IoT). Practically, problems related to remote sensors or devices are typically where edge computing and fog computing incorporate. Thermal comfort of urban open spaces is one of the most important topics in the field of edge and fog computing. It is necessary to fulfill the demands for more pleasant thermal comfort in urban planning and design new urban open spaces, as well as reviewing and improving the existing ones. The thermal comfort of urban open spaces is variable since it depends on climatic parameters and other influences, which are inconstant throughout the year, as well as during the day. Therefore, the prediction of thermal comfort is significant in order to enable planning of the usage time of urban open spaces. This research aims to develop an Improved Cuckoo Search (ICS) algorithm for forecasting physiological equivalent temperature (PET) values one hour ahead. Usually, the parameters of Cuckoo Optimization Algorithms (COAs) are kept constant, which may lead to efficiency reduction. To cope with this issue, a proper strategy for tuning the parameters is presented. Moreover, the generation of laid eggs is done by implementing the cross-over operator of a Genetic Algorithm (GA). Then, it is employed to train feed forward neural networks for PET prediction. Finally, the performance of the proposed algorithm is compared to the state-of-the-art; i.e., traditional COA and GA. Our simulation results demonstrate the effectiveness of the proposed algorithm for about a 93% compliance rate.

Health care in India has been ranked 112th position out of 190 countries by WHO (World Health Organization). The health care in this country falls behind due to high cost for treatment and medication. The Conventional robots used in the field of medical need high memory and computation power for decision making and task execution. Which in turn leads to a high budget in the construction of robots. To overcome this problem and to provide a solution for upgrading healthcare, we propose a new architecture which makes use of Commercial off-the-shelf (COTS) for cloud robotics. Then, the key forces behind the progress of cloud robotics are analyzed. This chapter mainly deals with the patient who needs first aid at times of emergency (based on the data from COTS product) and how the cloud robot stands by to help the patient. COTS help in tracking the health of individuals and these reports are stored either in the devices or on storage spaces. And the use of these products reduces the cost of the proposed model. As the data are stored in the cloud or on the device the robots do not require memory space for storing these data. Henceforth the memory requirement for the robot can be reduced which in turn the cost spent on the memory of the robot and its size.