to connect. The purpose of fog is to reduce the amount of data

Review: Security and Privacy Issues of Fog Computing Binara N. B. Ekanayake, Malka N. Halgamuge*, Azeem Mohammad School of Computing and Mathematics, Charles Sturt University, Melbourne, Victoria 3000, Australia Email: [email protected], [email protected], [email protected] Abstract -Internet of Things (IoT), devices, and remote data centers need to connect. The purpose of fog is to reduce the amount of data transported for processing, analysis, and storage, to speed-up the computing processes. The gap between, Fog computing technologies and devices need to narrow down as growth in business today relies on the ability to connect to digital channels for processing large amounts of data. Cloud computing is unfeasible for many internet-of-things applications, therefore fog computing is often seen as a viable alternative. Fog is suitable for many IoT services as it has enabled an extensive collection of benefits, such as decreased bandwidth, reduced latency, and enhanced security. However, Fog devices that are placed at the edge of the internet have met numerous privacy and security threats. This research presentation discusses the security and privacy issues of fog computing, through a comprehensive review of recently published literature in the area of security and privacy issues of fog computing, along with suggested solution for identified problems. Data extracted from 34 peer-reviewed scientific publications (2011 – 2017) were reviewed, leading to the identification of 49 different issues that were raised, in relation to fog computing. This study revealed a general agreement among researchers about the novelty of Fog computing, and its early stages of development, and identifies several challenges that need to be met, before its wider application and use reaches its full potential. Graphic Abstract Keywords Fog computing, edge computing, mobile edge computing, cloud computing, IoT, Security, Privacy, Big Data 1. INTRODUCTION Fog computing, also called fog networking or fogging, refers to a system of distributed computing infrastructure in which certain application processes are managed, by smart devices, at the edge of the network, although some data may still be managed in the cloud. Nonetheless, in most situations that involve IoT, cloud has limitations. Thus, Fog can be viewed as a middle layer between cloud and the hardware system. Having such a middle layer enables more efficient data processing, analysis and storage, and minimizes the amount of data that needs to be transported to the cloud. Fog computing was initially introduced as a concept to describe the technology that forms a link between remote data centers and Internet of Things (IoT) devices. In “fog computing” or “edge computing,” the system operates on network ends, instead of hosting and working entirely from a centralized cloud. It is designed to facilitate the storage of data, computing, and other services between end devices, mostly IoT devices, and cloud computing data centers [8]. Thus, the fog proposes a smart substitute that enables the processing of data locally, minimizing cloud involvement and enabling a smarter, more self-sufficient space between the data center and the network edge. Through a steep increase in the number of endpoint devices that communicate with the data center, it works as a latency that will escalate the network routine. Fog computing offers many other advantages, including more efficient real-time processing, rapid and affordable scaling, and local content and resource pooling. The ability to link the IoT with the existing Internet computing infrastructure has attracted considerable interest among academics, as well as of the IT industry. Fog supports (IoT) are growing rapidly, with the realization that the traditional cloud computing planning alone will never be enough. There are many applications that use fog computing currently, and it is likely that it will be in greater demand in the future. The areas that rely on this system include, connected vehicles, smart cities, smart grids, mobile healthcare systems, wireless sensor, actuator networks, connected manufacturing, connected oil, gas systems, autopilot vehicles, smart homes are names of a few [1]. In terms of mobile big data analytics, big data processing is still a new subject to the big data architecture in the cloud and mobile cloud. Fog computing is able to deliver flexible resources huge capacity data process system, without being hindered by the drawback of cloud’s high latency [34]. Another advantage of fog computing relates to data security which helps with confidentiality issues. Through the adoption of user behavior profiling, it can help to mitigate insider data theft attacks in cloud and combs both technologies that can be used for masquerade detection [35]. Fog computing has spiraled from cloud computing. However, fog computing has developed. several unique characteristics, the more important of which are; (i) Latency-sensitivity, location awareness and possessing edge location; (ii) being geographically spread out; (iii)involving a large number of nodes and a large-scale sensor networks; (iv)presenting real-time communications; (v) having heterogeneity; (vi) presenting interoperability and federation; (vii) having online analytics and interplay with the cloud [1]. At present, Fog computing is going through its early stages of development, and there are some challenging aspects that need to be addressed. Some of the concerns comprise issues that relate to privacy and security concerns such as; programming models and abstracts; fog architecture; IoT support; storage constraints, network, and computing; resource provisioning and management; and allocated fog computing centers [1]. For example, when we consider data protection, many IoT device applications have issues in fog computing. Handling of messages created from IoT devices, and sent to the closest fog nodes is a difficult task. To overcome this difficulty, the data is usually separated into few parts, and sent to numerous fog nodes, as there is a greater chance of assuring reliability. Since processed data is a compound that makes it more secure, nonetheless its challenge is to decrypt or encrypt data on IoT device as there are inadequate resources [6]. In fog computing security applications such as Smart Traffic Lights and Connected Vehicle, Wireless Sensor and Actuator Networks, IoT and Cyber-physical systems and Software Defined Networks, have some issues such as Verification at the smart meters which are installed in consumers’ house as well as numerous levels of the doorways. All the smart appliances and the smart meters have an IP address. A mischievous user can either report incorrect readings, interfere with the own smart meter or tricky IP addresses [5]. in the case of privacy in fog computing, there are some issues that have been identified as regarding as mobile device applications such as, places which have very weak surveillance and protection systems implemented into the fog node devices. They are local to the end user which makes it weak to security challenges. In contemplation of comprehending mischievous intentions become obtainable to fog computing framework such as; eavesdropping, data hijack and in-the-middle-attack, etc. [7]. A major focus of recent research has been on fog computing platform and applications. Researchers have briefly introduced Fog computing and after analyzing similar concepts of fog computing they have given a broader definition of fog computing [21]. Studies with surveys in fog computing concepts, applications, and issues that have been discussed the definitions of fog computing with similar concepts are given numerous types of issues that may find design and implement with fog computing systems [34]. Some authors have researched about research opportunities of the Fog and IoT, as their research indicates some future fog and its application in multiple industries and driving revolution through network operators like AT&T, IBM, and Huawei etc. [37]. New technologies such as
Fog computing have the potential to offer many benefits, privileges, conveniences, and efficiencies to the users. However, technological advances often present new problems, and one major concern is the protection of privacy and security of data. This study undertook a content analyses of the literature published in the area of security and privacy issues in fog computing, and the study is presented in four main parts: introduction, materials, and methods, results, discussion and conclusion. 2. MATERIALS AND METHOD The overarching research question for this study is: What exactly can be done to resolve security and privacy issues of fog computing? The data for this research is gathered from articles/research papers, on Fog computing security and privacy issues, published in scientific journals, during the 2011 to 2017 period. Table 1 provides a summary of the applications, of the issues that were addressed, and the respective techniques that were sugested as solutions, in some of the 34 publications that were included in this study. Table 1: Fog computing security and privacy issues using data extracted from 34 peer-reviewed scientific publications (2011 – 2017) No Author Consideratio n Application Issue Technique 1 Elkhati et al. (2017) [2] Fog Infrastructure Smart cities Home automation Data-driven industries • • • • • Cloud has the advantages of cost-effectiveness and scalability. However, it isn’t suitable for hosting all the applications. As a solution, off-loading some computations to the edge is proposed. The potential edge infrastructure is not well understood. There is no clarity on the types of applications that can be off-loaded. • • • Using micro-clouds which are collections of Raspberry Pis, to host a range of fog applications. They are particularly useful in networkconstrained environments. The startup latency, I/O overhead, serving latency and hosting capability have been experimentally tested for several different applications. 2 Hao et al. (2017) [3] Heterogeneity Universities Corporations Commonwealt h organizations Personal households. • • • • • There are heterogeneous of Fog nodes. There are no guarantees that the nodes will include comparable resources. Privacy and security issues are tied with the heterogeneity. They are mostly cast aside to accomplish interoperability and general functionality. When exchanging data to random devices, strict privacy policies and encryption create more complications. • • A flexible software architecture, incorporating different design choices and userspecified polices, is described. Also presented in a design of WM-FOG.A computing framework for fog environments that makes use of the proposed software architecture, and an evaluation of their prototype system. 3 Alrawais et al. (2017) [8] Secure and Efficient Protocols Smart grids Health care systems Wireless sensor networks Smart homes • • • Most of the existing protocols such as time synchronization use wireless packet transmissions. They aren’t suitable for resource-constrained IoT devices. Wireless transmissions and security computations utilize the major part of the energy budget. 4 Alrawais et al. (2017) [8] Authentication Smart grids Health care systems Wireless sensor networks Smart homes • • • Authentication in IoT has several problems related to scalability and efficiency. Traditional authentication methods are inefficient. Therefore, a secure, scalable, efficient, and user-friendly solution to cope with resourceconstrained IoT devices is required, • • The application of a lightweight encryption algorithm between fog nodes and IoT devices will improve the efficiency of the authentication process. Fog has the capacity to create an opportunity for authentication solutions in IoT devices, particularly wearable devices 5 Hu et al. (2017) [7] Privacy Mobile devices • Places which has very weak surveillance and protection are used to implemented the fog node devices. They are local to the end users. • Therefore, they are weak to security challenges. • In contemplation of comprehending mischievous intentions become obtainable to fog computing framework such as; eavesdropping, data hijack and in-themiddle-attack, etc. 6 Alrawais et al. (2017) [8] Updating Internet of things Devices Smart grids Health care systems Wireless sensor networks smart homes • • • Many IoT devices continue to be vulnerable to attacks. Management of security updates requires the design of remote software update capabilities. Vulnerable firmware may expose IoT devices to attacks that may not be protected by traditional security solutions such as firewalls. • • • • Fog computing can be helpful in finding a solution Fog can be used to identify vulnerabilities and to track firmware updates in IoT devices. However, updating billions of IoT devices is an unwieldy task. Security updating supply to IoT devices could be helped by the geodistribution characteristic of fog computing 7 Wen et al. (2017) [19] Reliability Tablets Smart Phones Desktop PCs • A large part of IoT applications is made of physical systems. Therefore, the assumptions made about fault and failure modes are weaker than those for Web-based applications. • IoT applications are vulnerable to crash and timing failures due to low-sensor battery power, high network latency, environmental damage etc. Also, the uncertainty caused by potentially unstable and mobile system components increases difficulties in • Therefore, methods are needed to measure an IoT application workflow’s reliability, and also apply ways of enhancing it. predicting and capturing system operation. 8 Wen et al. (2017) [19] Security Criticality Sensors Computer chips Communicatio n devices • • • Multiple sensors, computer chips, and communication devices are integrated to enable overall communication, in the IoT environment, There may be multiple components in a given service, with each deployed in its own geographic location, This makes each a separate attack target. Fog nodes can be easily attacked, especially those in network enabled IoT systems, where attack vectors can include human-caused sabotage of network infrastructure, malicious programs provoking data leakage, or even physical access to devices • • Accurate evaluation of the security and risks to obtain a holistic measure of security and risk susceptibility, is critical. This becomes challenging when workflows are changing and adapting to runtime. 9 Wen et al. (2017) [19] Dynamicity IOT Services • • Software upgrades via fog nodes or the frequent join-leave behavior of network objects, will cause changes to internal properties and performance, which in turn can change the overall workflow execution pattern. Software and hardware aging is a concern with handheld devices because that too will change workflow behavior and device properties • There is a need for automatic and intelligent reconfiguration of the topological structure and assigned resources within the workflow, and importantly, that of fog nodes. • Changes in application performance are due to their transient and/or short-lived behavior within the system 10 Wen et al. (2017) [19] Fault Diagnosis and Tolerance IOT Services • Scaling a fog system increases the probability of failure • Certain software bugs or hardware faults that may be harmless at smaller scale or in testing environments, such as stragglers, can be devastating on system performance and reliability. • Different fault combinations may occur at the scale, heterogeneity, and complexity. • Developers should incorporate redundant replications and usertransparent, faulttolerant deployment and execution techniques in orchestration design 11 Xiao et al. (2017) [24] Wireless Networking (Vehicular Fog Computing) Vehicles • Because of changes to the connectivity of vehicles and frequent changes in the network topology, the reliability of vehic
le networking still causes problems, although a recent field test showed that 802.11p performs reliably most of the time. • The existing performance analysis of D2D-based vehicular networking is based on simulation. • Also, there are concerns that the cellular networks may not fully cover all the urban and suburban areas and the bandwidth resources of cellular networks are limited. • In vehicular fog computing scenarios where D2D and WLAN-based approaches co-exist, can be used. The coscheduling must take into account the performance requirements (i.e. bandwidth, latency) and the available capacity. Coscheduling mechanisms have still not gained sufficient attention. 12 Xiao et al. (2017) Application Provisioning Vehicles • Vehicular and latency-sensitive • Huang et al. [26] proposed an adaptive [24] (Vehicular Fog Computing) • • mobile applications can be deployed on central cloud, cellular fog nodes and/or vehicular fog nodes Pham et al. [25] discussed task scheduling between rented cloud nodes and owned fog nodes, and proposed a heuristic-based algorithm as a way of balancing between the make-span and the monetary cost of cloud resources. Mobility of fog nodes pose new challenges such as those arising from the simultaneous mobility of both vehicular fog nodes and their data sources and users. Another arises from the complexity of coordinating the scheduling of computing and communications resources. • content reservation scheme to reserves the resources on cloud and fog nodes for real-time streaming to mobile devices. This takes account of the mobility of mobile devices. Lin et al. [27]’s idea was to develop a Cloud-fog that utilizes the idle machines of game players. It depends on the organization of fog nodes for rendering game videos and streaming them to nearby players. 13 Xiao et al. (2017) [24] Security and Privacy Vehicles • • Another avenue of security and privacy challenge involves the mobility of fog nodes and the dynamic vehicular network topology. The distributed vehicular fog nodes serve as gateways to the hybrid cloud consisting of fog nodes and central cloud. • As a way of protecting against malicious attacks, Mtibaa et al. [28] implemented Honey Bots to detect and track the activities of malicious communications in D2D network. Public key infrastructure [29] and DiffieHellman key exchange [30] have been elaborated to enhance the security of authentication in • • A hacker who gets access to any of the fog nodes, can send malicious messages and illegitimate commands that can seriously harm the reliability of the network services Attackers can also duplicate the personal data of the clients by hacking into the vehicular fog nodes, seriously threatening the clients’ privacy. • smart grid networks. However, the high expansion rate of the vehicle fog platform may continue to cause more security problems in the future. • More effective encryption methods and powerful middleware need to be developed for security-ware of fog computing to address such challenges, 14 Niranjanamurthy et al. (2016) [12] Compute/ Storage limitation Wireless access points • There are attempts currently to expand storage capabilities with devices that are smaller, have better energy-efficiency and power. • For example, a present day phone has more power than many of the desktops used 15 years ago. • For nonconsumer devices more and more improvements are being made. 15 Niranjanamurthy et al. (2016) [12] Management Wireless access points Smart traffic lights • IoT/ubiquitous computing nodes and applications running on top must be put in place and configured to operate as required, to set up communication routes across end nodes. • FOG be dependent much on distributed (scalable) management mechanisms. • Possibly there are billions of small devices that need such configuration. • As fog and asymptotic declarative configuration techniques become more common, it is unlikely that there will be complete control achieved of the whole. • They are to be examined at this unparalleled scale. 16 Niranjanamurthy et al. (2016) [12] Standardizatio n Wireless access points Smart traffic lights Smart cities • Presently, there are no standardized mechanisms. • The availability of the different members of the network (terminal, edge point…) could be announced, so that others can send their software to be run. 17 Niranjanamurthy et al. (2016) [12] Accountability /Monetization Wireless access points Smart traffic lights • The facility for users to share their spare resources to host applications will help in the development of new business models around the fog concept. • It needs the creation of an incentive scheme. •Inducements can be financial. (e.g. unlimited free data rates). • It is difficult to regulate if a given device is hosting a section (droplet) or not at a given time, in the absence of a central controlling entity in the fog. 18 Niranjanamurthy et al. (2016) [12] Discovery/Syn c Wireless access points Smart traffic lights • For applications using devices will need agreed, central point (e.g. if there are too few peers in the storage application to establish an upstreamǁ backup) 19 Khalid (2016) (14) Network fortification Smart home/cities Smart meters connected vehicles Generous scale remote sensor networks • Because of the extensive use of wireless networking in fog computing, remote framework or wireless network security is a major concern in fog computing. • The examination range of wireless networks can be subjected to attacks such as jamming and sniffing. • Fog nodes are located at the edge of Internet, which certainly passes on an overpowering load to the network management. • As cloud servers are scattered all through the framework/network edge, there is a high cost associated with their maintenance. • • • The control of software defined networks can enhance the execution and management Techniques to introduce adaptability of network and lessening expenses, applicable to fog computing. Application of SDN technique in fog computing will offer fog computing security novel capabilities and prospects. 20 Khalid (2016) (14) Access Control Smart home/cities Smart meters connected vehicles Generous scale remote sensor networks • Standard access control is ordinarily handled in the same trust region. • For out-sourced data in cloud computing, the access control is, in general, cryptographically realized. • Symmetric key based • Some open-key based approaches for action are proposed as an effort to fulfill finegrained access control. • Other authors [eg: Yu et al, 2010] suggest a fine-grained data access control preparation generated course of action is not versatile in key management. on attribute-based encryption (ABE). • Furthermore, to support secure joint exertion and interoperability among various resources, Dsouza et al, (2014) proposed a policy-based resource access control in fog computing, 21 Kumar et al. (2016) [22] Network Security Smart metering system Smart wearable devices Smart cities • • • Fog computing is affected by attacks like sniffer, spoofing, jamming etc. Usually such attacks are targeted between the fog node and the centralized system. Fog nodes are at the edge of the network. Therefore, it increases the burden for the network manager. • • SDN (Software Defined Networking) can be used as an approach for network managers to work at the low level of abstraction for network services. It can help in management, increase scalability of network as well as reduce costs of fog computing. • To watch the traffic, we can use Intrusion Detection System and Network Monitoring. To prevent attack Prioritization system and Traffic Isolation can be used by shared resources, Network resource access control system helps to get access control on SDN (Open Control), Network Sharing System can help the fog node router to be open to guests considering the secu
rity issues as well. 22 Kumar et al. Data Security Smart metering • It is difficult to maintain data • To provide data verifiability, (2016) [22] system Smart wearable devices Smart cities Integrity, as data may be lost or be modified. • The data uploaded to the fog node can also be used by a third party. confidentiality and Integrity combination of searchable encryption techniques and homomorphic encryption can be used. • These techniques will be useful for ensuring that clients do not store data on untrusted servers. 23 Kumar et al. (2016) [22] Access Control Smart metering system Smart wearable devices Smart cities • Access control is significant as a tool to provide system’s security and ensure user privacy. Usually access control is labeled to the same domain, but better to implement cryptographically, because of the distributive nature of cloud computing • • Many solutions have been proposed for these problems. One of them by Yu et al recommends that the access control is based on Attributebased encryption (ABE). • Others proposed solutions are theorybased with suggestions for policy based access control mechanism is applied to handle the heterogeneous nature of fog computing. 24 Petac et al. (2016) [23] Security Routers Switches IP based video cameras • Security of stored data is one of the major concerns in the fog computing security area. • All data, in this environment, is stored within a third party. • This makes the implementation of traditional security solutions A more difficult proposition. The use of cryptographic methods for data storage, although is better in • • Proposed, Adaptive Fog Computing Node Security Profile (AFCNSP) based on security Linux solutions. In the case of fog nodes, the decision about what kind of security is necessary is important. Without authorization, a fog node application cannot access the services, the data and the network of other application • Simplified terms of security, creates problems for the users, because the latter will not have any control over their own data. Mandatory Access Control Kernel –, Discretionary Access Control – DAC, Cynara and net filter are Linux kernel security modules that protects data and process interactions from malicious manipulation by using a set of custom mandatory access control rules. 25 Wang et al. (2015) [4] Geographical distribution Wireless Sensor and Actuator Networks Decentralized Smart Building Control Software Defined Networks IoT and Cyber physical systems • A huge number of nodes and sensors are needed In Fog computing environment. • It helps to release the restrictions of bandwidth for advanced communicational speed in the real-time interactions and network. 26 Lee et al. (2015) [6] Data Protection Internet of things devices Smart watches • Messages which are created from IoT devices were sent to the closest fog nodes. • To manage many data on IoT devices are challenging. • To handle the data, it is separated into few parts as well as sent to numerous of fog nodes. •Without revealing the substances of data should be analyzed. • The reliability of data must be assured, while it’s been distributed. • Processed data is compound. • It is challenging to decrypt or encrypt data on IoT device as there are inadequate resources. 27 Lee et al. (2015) [6] Data Management Internet of things devices Smart watches • To make it challenge to find data’s location, the fog nodes are geographically allocated. • In the other areas also, the users require equal services. • It is very challenging to find out whether the node delivers the equal service. • By having replicated files, it will cause a waste of resources. • Result from a wrong approach by a mischievous user fog nodes got distributed on security issues of personal information. 28 Yue Shi et al. (2015) [11] Attack Detection Mobile devices Wi-Fi access points • Fog computing provides new opportunities to detect unusual behavior or to identify malicious attacks. • A detection system can be signature- or anomaly-based. • A newly detected pattern can be checked against existing or possible patterns. • Cloudlet mesh architecture through the collaboration of cloudlet members to monitor and detect malware, malicious attacks, and other threats, is proposed. • Such a collaborative intrusion-detection technique will be applicable between fog nodes to monitor IoT environments and their surroundings. 29 Stojmeno System Smart grids • Man-in-the-middle attack • Complete avoidance and vic et al. (2015) [13] security Smart traffic lights in vehicular networks Software defined networks • • • • Fog computing is typically vulnerable to the man-in-the-middle attack. Gateways serving as Fog devices may be altered or replaced completely, in such attacks. For an example, Star Bar or KFC customers being connected to mischievous access points which provide misleading service set identifiers as public legitimate ones. once the attackers take control of the gateways, users’ private communications will be stolen. defending against manin-the-middle attacks are difficult tasks. • However, building an anti-tampering mechanism in the Fog device could be done as a potential solution. 30 Stojmenovic et al. (2015) [13] Authentication and authorization Smart grids Smart traffic lights in vehicular networks Software defined networks • The connection between Fog and Cloud is fragile, and as such can be easily disrupted. In such instances, authentication of users could be a problem as it is deployed on the Cloud server. • A new mechanism is introduced for user authentication when there is no connection to the Cloud server. It is known as StandAlone Authentication (SAA). 31 Yi et al. (2015) [17] Secure Data Storage Wireless sensor networks Smart home/cities Smart meters • Just like in cloud computing security threats are there in fog computing too because user data is outsourced and user’s control over data is handed over to fog node. • Maintenance of data integrity is a problem because the outsourced data may be lost or changed. • Furthermore, the data may be vulnerable to abuse by unauthorized parties • As a means of countering such threats, for auditable information storage service there have been proposals. •Techniques being searchable encryption and homomorphic encryption can be used. • Public auditing for data stored in cloud, which relies on a third-party auditor (TPA), using random mask and technique homomorphic authenticator has been proposed to for fraudulent ensure privacy. 32 Yi et al. (2015) [17] Data Privacy Wireless sensor networks Smart home/cities Smart meters • Privacy-regulating algorithms in the fog network are run between the fog and cloud, while those algorithms at the end devices are usually resource prohibited. • Sensitive data generated by sensors and end devices are usually collected by fog nodes at the edge. • Privacy- assurance aggregation at the local gateways without decryption can be enhanced by techniques such as homomorphic encryption. 33 Chiang et al. (2015) [20] Trustworthiness and security 5G, home/ personal networking Internet of Things • Fog may be useful in enhancing security in some cases. However, there is also the possibility of new security challenges, at times. • Since it is rather easier to hack into client software, perhaps consideration of security at hardware level on devices could be worthwhile. 34 Yi et al. (2015) [21] Security and Privacy Health Data Management Smart Home Smart Grid Smart Vehicle •Admittedly, ensuring security and privacy is one of the biggest challenges in every stage of fog computing platform design. •These problems can be overcome by applying intrusion detection system, and access control which require assistance from every layer of the platform. 35 Yi et al. (2015) [21] Network Management Health Data Manageme
nt Smart Home Smart Grid Smart Vehicle • • One of the major issues in fog computing is network management which can be tackled by using SDN and NFV techniques. However, the integration of SDN and NFV into fog computing is no easy task. • The need to re-design the north-bound and south-bound, the east-west-bound APIs to include necessary fog computing primitives pose many challenges, 36 Yi et al. (2015) [21] Fight with Latency Health Data Management Smart Home Smart Grid Smart Vehicle • since fog computing is aiming delaysensitive applications and services, high inactivity will have negative effects on user satisfaction and experience. • Many factors would introduce high prospective into service or application implementation on fog computing platforms. 37 Yi et al. (2015) [34] Fog networking Wireless network virtualization Privilege traffic reservation Frequency hopping communicatio n • Located at the edge of Internet, fog network is heterogeneous. Fog network’s task is to connect all its components. However, managing such a network, maintaining connectivity and providing services, especially in the scenarios of the Internet of Things (IoT), is a challenging task. •Software-defined networking (SDN) and network function virtualization (NFV) are two of the emerging techniques that have been proposed for use to create flexible and easy maintaining network environment. 38 Yi et al. (2015) [34] Connectivity (Quality of Service) ad-hoc wireless sensor networks Smart phones • Opportunities for cost-cutting, data trimming and connectivity expansion are provided by network relaying, partitioning and clustering. For example, due to the coverage of richresource fog nodes an ad-hoc wireless • Work [31] proposes an online AP association strategy that not only achieves a minimal throughput, but efficiency in computational overhead. Similarly, the selection of fog node from end user will heavily impact sensor network can be partitioned into several clusters (cloudlet, sink node, powerful smartphone, etc.). the performance. • A subset of fog nodes can be selected as relay nodes for optimization goals of maximal availability of fog services limited to a certain area or a single user. It can also include constraints such as delay, throughput, connectivity, and energy consumption. 39 Yi et al. (2015) [34] Reliability (Quality of Service) Clustering computing Grid computing Cloud Sensor networks •Periodical check-pointing to resume after failure, rescheduling of failed tasks or replication to exploit executing in parallel can improve reliability. But check pointing and rescheduling may not suit the highly dynamic fog computing environment since there will be latency period, and adaptation to changes may be slow. •Replication appears to be more promising, but relies on multiple fog nodes working together. 40 Yi et al. (2015) [34] Capacity (Quality of Service) Cloud Sensor networks • Capacity has two folds: network bandwidth and storage capacity • To achieve high bandwidth and efficient storage utilization, it is important to investigate how data are placed in fog network. In computation, data locality is very important. There are similar works in the context of cloud, and sensor networks • This is an issue that can cause problems in fog computing. For example, a fog node may need to compute on data that is distributed in several • The problem can be solved by leveraging user mobility pattern and service request pattern to place data on suitable fog nodes to either minimize the cost of operation, the latency or to maximize the throughput. • Fog computing offers the possibility of dynamic data placement and large overall capacity. Therefore, it may need to redesign search engine which can process search query of content scattered in fog nodes • It would be of interest to redesign cache on fog node to exploit temporal locality and broader nearby nodes. Since computation requires the finish of data aggregation, it can delay the services. coverage to save network bandwidth and reduce delay, while there is existing work of cache on end device and cache on edge router. 41 Yi et al. (2015) [34] Delay (Quality of Service) Streaming Complex event processing • Latency-sensitive applications, such as streaming mining or complex event processing, are typical applications which need fog computing to provide real-time stream processing rather than batch processing •According to Hong et al an opportunistic spatiotemporal event processing system can be used to address the latency issue. This system predicts future query region for moving users, and the event processing is done early to make information available when the user reaches a different location. 42 Yi et al. (2015) [34] Interfacing and programming model Developers Internet applications • To ease the developers’ efforts to port their applications to fog computing platform, unified interfacing and programming model are needed, because • Application-centric computing will be a useful fog computation model. Here, the components will allow suitable optimizations and application-aware for different kinds of applications. • It is hard for a developer to put together heterogeneous resources, and hierarchical dynamic • Hong et al. [33] suggest a high-level programming model for future Internet applications with on-demand scaling. They are large-scale geo-spatially distributed and latency sensitive. • However, their system depends on a tree-based network hierarchy where fog nodes have fixed locations. Also needed are more general schemes for diverse networks with fog nodes of dynamic mobility. to build well-matched applications on diverse platforms. 43 Yi et al. (2015) [34] Computation Offloading Mobile devices • Dealing with dynamics is one of the main challenges in offloading in fog computing, The dynamics are three fold 1) radio/wireless network access 2) nodes in the fog network 3) resources in the fog. • The combining of fog and cloud actually would give a three-layering construction: device-fog cloud. However, computation offloading in such infrastructures can lead to both new challenges and opportunities. • To choose the type of granularity for offloading at different hierarchy of cloud and fog; how to make offloading conclusions to adjust active changes in resources, fog devices and network, etc. also how to partition application to offload on cloud and fog with dynamism are the major questions. 44 Yi et al. (2015) [34] Accounting, billing and monitoring Cloud service providers End users Internet service providers • The success of Fog computing enterprise depends on a sustainable business model Following parties are required in fog computing service: • Cloud service providers • End users User Incentives • “Join Fog computing with private local cloud at the edge” is an innovative business model in this field. • Local private clouds with computation and • • • • • Internet service providers or wireless carriers Therefore, there are many issues to be resolved for the introduction of a “Pay-as-you-go” system. For example, in billing, decisions have to be made on how to set the price for different resources and how to set the fraction of the payment to go to different parties of fog Methods of accounting and monitoring the fog in different granularity are required for enforcing such pricing models. Another interesting feature will be the ability to do dynamic do pricing in fog computing services to maximize revenue and utilization, just similar to traditional industries do in airline ticketing, car rental or hotel bookings. • storage capacity., can be deployed at the edge of Internet, although private cloud is aiming at service to private parties only. It will also be possible to lease spare computation and storage facilities to fog service provider for a fee which will also help in reducing the costs. 45 Yi et al. (2015) [34] Intrusion detection •
There are new opportunities to investigate how fog computing can help with the centralized cloud side and the interruption recognition on both client side. • However, high mobility fog computing environment, largescale, implementing intrusion detection in geo-distributed pose some challenges. 46 Yi et al. (2015) Privacy Cloud • Internet users are anxious about the risk of privacy • Privacy- preserving [34] Smart grid Wireless network Online social network • • leakage (data, location or usage). There are many suggested techniques for preserving privacy for different scenarios such as cloud, smart grid, wireless network and online social networks In fog network, Privacypreserving algorithms can be located between the cloud and fog since storage and computation are acceptable for both sides while at the end devices those algorithms are usually resourceprohibited. • Data is produced by end devices and sensor, fog node at the edge generally gather data from there. aggregation at the local gateways without decryption. can be facilitated by techniques like homomorphic encryption. For aggregation and statistical queries, differential privacy can be applied to ensure non-disclosure of an arbitrary single entry in the data set, without affecting privacy. 47 Yi et al. (2015) [34] Access control smart devices cloud • Because of its ability ensure security, access control is a reliable tool on smart devices, and cloud, • The way to design control spanning client-fog-cloud, to meet the goals and resource constraints at different levels, is another issue in fog computing 48 Stojmenovic (2014) [5] Security Smart Traffic Lights and Connected Vehicle Wireless Sensor and Actuator Networks IoT and Cyber- • • • Verification at the smart meters which are installed in consumer’s house as well as numerous levels of the doorways. All the smart appliances and the smart meters has an IP address. A mischievous user can either report incorrect readings, interfere with physical systems Software Defined Networks the own smart meter or tricky IP addresses. 49 F. Bonomi (2012) [9] F. Bonomi (2011) [10] Programmability Programmers Advertising filed Entertainment and other applications • The control of application lifecycle is one of the challenges in cloud environments • . As there are many small functional units (droplets) distributed over many devices, they require the correct placing of the abstractions • It may not be necessary for programmers to deal with these issues. 3. RESULTS Table 2 shows the considerations and applications according to the year. Table 2: Considerations and applications according to the year Year Consideration Application 2017 Fog Infrastructure • Smart cities • Home automation • Data-driven industries Heterogeneity • Universities • Corporations • Commonwealth organizations • Personal households. Secure and Efficient Protocols • Smart grids • Health care systems • Wireless sensor networks • Smart homes Authentication • Smart grids • Health care systems • Wireless sensor networks • Smart homes Privacy • Mobile devices Updating Internet of things Devices • Smart grids • Health care systems • Wireless sensor networks • Smart homes Reliability • Tablets • Smart Phones • Desktop PCs Security Criticality • Sensors • Computer chips • Communication devices Dynamicity • IOT Services Fault Diagnosis and Tolerance • IOT Services Wireless Networking (Vehicular Fog Computing) • Vehicles Application Provisioning (Vehicular Fog Computing) • Vehicles Security and Privacy • Vehicles 2016 Compute/ Storage limitation • Wireless access points Management • Wireless access points • Smart traffic lights Accountability/Monetization • Wireless access points • Smart traffic lights Standardization • Wireless access points • Smart traffic lights • Smart cities Discovery/Sync • Wireless access points • Smart traffic lights Network fortification • Smart home/cities • Smart meters connected vehicles • Generous scale remote sensor networks Access Control • Smart home/cities • Smart meters connected vehicles • Generous scale remote sensor networks • Smart metering system • Smart wearable devices Network Security • Smart metering system • Smart wearable devices • Smart cities Data Security • Smart metering system • Smart wearable devices • Smart cities Security • Routers • Switches • IP based video cameras 2015 Geographical distribution • Wireless Sensor and Actuator Networks • Decentralized Smart Building Control • Software Defined Networks • IoT and Cyber physical systems Data Protection • Internet of things devices • Smart watches Data Management • Internet of things devices • Smart watches Attack Detection • Mobile devices • Wi-Fi access points System security • • • Smart grids Smart traffi c lights in vehicular networks Software d efined networks Authentication and authorization • • • Smart grids Smart traffic lights in vehicular networks Software defined networks Secure Data Storage • Wireless sensor networks • Smart home/cities • Smart meters Data Privacy • Wireless sensor networks • Smart home/cities • Smart meters Trustworthiness and security • 5G, home/ personal networking • Internet of Things Security and Privacy • Health Data Management • Smart Home • Smart Grid • Smart Vehicle • Cloud • Wireless network • Online social network Network Management • Health Data Management • Smart Home • Smart Grid • Smart Vehicle Fight with Latency • Health Data Management • Smart Home • Smart Grid • Smart Vehicle Fog networking • Wireless network virtualization • Privilege traffic reservation • Frequency hopping communication Connectivity (Quality of Service) • ad-hoc wireless sensor networks • Smart phones Reliability (Quality of Service) • Clustering computing • Grid computing • Cloud • Sensor networks Capacity (Quality of Service) • Cloud • Sensor networks Delay (Quality of Service) • Streaming • Complex event processing Interfacing and programming model • Internet applications • Developers Computation Offloading • Mobile devices Accounting, billing and monitoring • Internet service providers • Cloud service providers • End users Access Control • Smart devices • cloud 2014 Security • Smart Traffic Lights and Connected Vehicle • Wireless Sensor and Actuator Networks • IoT and Cyber-physical systems • Software Defined Networks 2011 Programmability • Programmers • Advertising filed • Entertainment and other applications Table 3: Category of Applications, and their description: Data from 2015 to 2017 Category Description Smart Cities Smart cities Networks Wireless sensor networks, Wireless access points, Wireless Sensor and Actuator Networks, Software Defined Networks, Wireless networks, Wireless network virtualization, Wi-Fi access points, Generous scale remote sensor networks, 5G, home/ personal networking, ad-hoc wireless sensor networks Grids Smart grids Grid computing Healthcare Healthcare systems Health Data Management Smart homes/Households Smart homes Personal households / Home automation Decentralized Smart Building Control Mobile Devices Mobile devices Tablets / Smart Phones Vehicles Vehicles, Smart meters connected vehicles, Smart Vehicle Traffic lights Smart traffic lights in vehicular networks, Smart traffic lights, Privilege traffic reservation Meters Smart meters Smart metering systems Wearable devices Smart wearable devices Smart watches IOT
IOT Services Internet of Things Internet applications IOT and Cyber physical systems Cloud Cloud Other Data-driven industries, Universities, Corporations, Common wealth organizations, online social network, Desktop PCs, Sensors, Computer chips, Communication devices, Routers, Switches, IP based video cameras, Frequency hopping communication, Clustering computing, Streaming, Complex event processing, Developers Table 4 is obtained from the data from articles that published between 2015 to 2017. Within this table, only the application headings have been mentioned, and the table above explains it in a broader application. Table 4: Number of applications has been used in each year (2015 to 2017) Application Number of applications Smart Cities Year Number 2017 2 2016 4 2015 2 Total 8 Networks Year Amount 2017 3 2016 7 2015 13 Total 23 Grids Year Amount 2017 3 2015 6 Total 9 Healthcare Year Amount 2017 3 2015 3 Total 6 Smart homes/Households Year Amount 2017 5 2016 2 2015 6 Total 13 Mobile devices Year Amount 2017 3 2015 3 Total 6 Vehicles Year Amount 2017 3 2016 2 2015 3 Total 8 Traffic lights Year Amount 2016 4 2015 3 Total 7 Meters Year Amount 2016 3 2015 2 Total 5 Wearable devices Year Amount 2016 3 2015 3 Total 6 IOT and services Year Amount 2017 3 2015 4 Total 7 Cloud Year Amount 2015 4 Total 4 0ther Year Amount 2017 8 2016 3 2015 6 Total 17 Total no of applications 119 Figure 1, based on the data from Table 4, indicates that 19% of the issues were found from network regarding applications, while smart homes/ household’s applications accounted for 11% of the issues. Figure 2shows the percentages of issues in fog computing that have either been/not been suggested a solution. Of the total number of 49 issues, only 32 (65%) have been given solutions or techniques. 4. DISCUSSION This study is a review of 34 research papers, that were published during 2011 – 2017, on the subject of security and privacy issues associated with Fog computing. An important observation is the relatively low number of articles published on Fog computing issues during the said period, highlighting the need for more research done in this important area. As seen from the distribution of issues that were addressed (Figure 1), 19% of the research related to network regarding applications. Smart homes/ household’s applications ranked second in frequency, with 11% of papers related to research in this area. Thus, most of the issues of research interest related to the network section, and the reason could be because fog computing is still in its early stages. All in all, there were 49 issues that were covered in the collection of publications that were reviewed. Examples of some of the highlighted issues include, lack of clarity on the types of applications (2) and problems of scalability and efficiency (8). The major concern, however, as indicated by the majority of papers relate to privacy and security of data (eg: 24, 28 & 29). Some authors drew attention to the vulnerability to both attacks as well as crashes (8, 19). However, it has also been suggested that Fog computing itself may be capable of offering solutions to such attacks (8). Many have drawn attention to the need for configuration and topological structural changes to Fog nodes (19, 24, 26, 27). The need for, and current efforts to expand storage capabilities have also been discussed (12). Of the 49 issues raised, 32 (65%) have been given a solution or a suggested technique, while there has been no offer a solution for the rest. The fact that research has failed to suggest any solutions for onethird of the identified problems, once again, highlights the need for more research in this area. A similar observation was made by Baccarelli et al. [36] who discussed the energy-efficient networking and computing architectures, after carrying out a statistical search on Google, using keywords such as Internet of Everything (IOE) and Fog. The search was limited to the research contributions published during 2011 to 2016, and the authors drew attention to two main findings. Firstly, there has been a move towards a combination of the Fog and Internet of Everything paradigms. Secondly, as for the actual integration of the Fog and Internet of Everything pillar paradigms, there is still no up-to-date research or technical contribution that has been published. An overview of research opportunities related to the Fog and IoT, was the focus of Chiang et al. [37]. From their analysis, the authors concluded that in the future, fog will be applicable in multiple industries and creating a revolution through the entire industry, including network operators like AT&T, Network equipment vendors like Huawei, System integrators like IBM, and end user experience providers like Toyota etc. Considering the studies that were analysed in the current review, summation of the papers (Table 1) shows that authors have discussed a multitude of issues regarding fog computing, each paper discussing one or two aspects of a particular issue, some taking the extra step to suggest solutions for the identified problems, while others have limited to a discussion of the issue only. For example, Yi et al. [21] have discussed fog computing platform and applications. They have briefly introduced fog computing, and following the analyzes, several concepts on fog computing have emerged as the authors gave a broader definition of fog computing. In the paper, they have discussed the challenges and design goals for fog computing. Then they designed an implementation of prototyping platforms for fog computing. Finally, they tested prototypes and platforms in smart home applications. Yi et al. [34] has done a survey of fog computing concepts, applications and issues. The survey has been discussed about the definitions of fog computing with similar concepts and has given numerous issues that may arise when designing and implementing fog computing systems. The authors have discussed the challenges, new opportunities and techniques in fog computing and issues such as quality of service, privacy, security, resource management and interfacing, have been emphasized. As they indicate, fog computing will evolve with rapid development in underlying Mobile cloud, NFV, radio access techniques, IoT, SDN, VM and edge devices. 5. CONCLUSION In this study, we have reviewed and analyzed the data obtained from 34 published research articles (papers regarding issues in fog computing) from 2011 to 2017, and evaluated 49 distinct issues that have been addressed, relating to fog computing. From the 49 issues, 32 issues have been provided with a solution technique and rest of the17 issues have not been given a solution or a technique, suggesting the need for greater attention of researchers to these areas. This study also found that 19% of the fog computing related to applications issues of through networks. Additionally, 11% of the issues were found from smart homes/ household’s applications as they are the second biggest issues in those articles. Consequently, it’s clear that most of the issues are regarding the network section are from the application of fog computing. When reviewing the research articles, it was quite clear that most of the studies focused on security or privacy and there were no issues that were concerned with networks applications. Furthermore, past research papers too, have not shown much concern about issues of fog computing applications, in a comprehensive manner. All the authors have only discussed a particular aspect of fog computing, but no one ever discusses about all the issues in a single document. In this research paper, we have attempted to fill that gap, and highlight the areas that need attention in the future. It is evident from the review that Fog computing is still a new area today, and is still not well understood or researched, despite the significant role it is likely to play in the future. There are challenges in establishing solutions to the issues that are raised, but finding solutions require urgent attention. Author Contribution B.N.E. and M.N.H. conceived the study idea and developed the analy
sis plan. B.N.E. analysed the data and wrote the initial paper. M.N.H. helped to prepare the figures and finalizing the manuscript. All authors read the manuscript. 6. REFERENCES [1] S. Chen, T. Zhang and W. Shi, “Fog Computing”, IEEE Computer Society, 2017. [2] Y. Elkhatib, B. Porter, H. B. Ribeiro, M. FatenZhani, J. Qadir and E. Rivière, “On Using Micro-Clouds to Deliver the Fog”, IEEE Computer Society, 2017. [3] Z. Hao, E. Novak, S. Yi and Q. Li, “Challenges and Software Architecture for Fog Computing”, IEEE Computer Society, 2017. [4] Y. Wang, T. Uehara and R. Sasaki, “Fog Computing: Issues and Challenges in Security and Forensics”, IEEE Computer Society, 2015. [5] I. Stojmenovic and S. Wen, “The Fog Computing Paradigm: Scenarios and Security Issues”, Proceedings of the 2014 Federated Conference on Computer Science and Information Systems pp. 1–8, 2014. [6] K. Lee, D. Kim, D. Ha, U. Rajpu and H. Oh, “On Security and Privacy Issues of Fog Computing supported Internet of Things Environment”, IEEE Computer Society, 2015. [7] P. Hu, H. Ning, T. Qiu, H. Song, Y. Wang and X. Yao, “Security and Privacy Preservation Scheme of Face Identification and Resolution Framework Using Fog Computing in Internet of Thing”, IEEE Computer Society, 2017. [8] A. Alrawais, A. Alhothail, C. Hu and X. Cheng, “Fog Computing for the Internet of Things: Security and Privacy Issues”, IEEE Computer Society, 2017. [9] F. Bonomi, “Fog computing and its role in the internet of things.” Proceedings of the first edition of the MCC workshop on Mobile cloud computing. ACM, 2012, pp. 13-16). [10] F. Bonomi, “Connected vehicles, the internet of things, and fog computing,ǁ in The Eighth ACM International Workshop on Vehicular InterNetworking (VANET), Las Vegas, USA, 2011. [11] Y. Shi, S. Abhilash, and K. Hwang, “Cloudlet Mesh for Securing Mobile Clouds from Intrusions and Network Attacks,” Proc. 3rd IEEE Int’l Conf. Mobile Cloud Computing, Services, and Eng., 2015, pp. 109–118. [12] M.Niranjanamurthy, P.B.Kavitha, K.Priyanka ,S.N.Vishnu, “Research study on fog computing for secure data security,” International Journal of Science Technology and Management, Vol. No.5, Special Issue No.(01),February 2016, ISSN2394-1537. [13] I.Stojmenovic, S.Wen, X.Huang and Hao Luan, ” An overview of Fog computing and its security issues”, Concurrency and Computation: Practice and Experience, Volume 28, Issue 10 July 2016 Pages 2991–3005. [14] K.A. Fakeeh, “Privacy and Security Problems in Fog Computing”, Communications on Applied Electronics (CAE) – ISSN : 2394-4714, Foundation of Computer Science FCS, New York, USA, Volume 4– No.6, March 2016. [15] C.Dsouza, G.Joon Ahn M.Taguinod, “Policy-Driven Security Management for Fog Computing: Preliminary Framework and A Case Study”, IEEE Computer Society, 2014. [16] S.Yu, C.Wang, K.Ren andW.Lou, Achieving Secure, Scalable, and Fine-grained Data Access Control in Cloud Computing, IEEE INFOCOM 2010. [17] S.Yi., Z. Qin., Q.Li. (2015) Security and Privacy Issues of Fog Computing: A Survey. In: Xu K., Zhu H. (eds) Wireless Algorithms, Systems, and Applications. WASA 2015. Lecture Notes in Computer Science, vol 9204. Springer, Cham. [18] C. Wang, Q. Wang, K. Ren, W. Lou: Privacy-preserving public auditing for data storage security in cloud computing. In: INFOCOM. IEEE (2010). [19] C. Wen, R. Yang, P. Garraghan, T. Lin, J. Xu, M. Rovatsos, “Fog Orchestration for Internet of Things Services”, IEEE Computer Society 2017. [20] M. Chiang, A.L. Doty, “Fog Networking: An Overview on Research Opportunities”, Electrical Engineering Princeton University, 2015. [21] S. Yi, Z. Hao, Z. Qin, and Q. Li, “Fog Computing: Platform and Applications”, 2015 Third IEEE Workshop on Hot Topics in Web Systems and Technologies. [22] P. Kumar, N. Zaidi and T. Choudhur, “Fog Computing: Common Security Issues and Proposed Countermeasures”, Proceedings of the SMART -2016, IEEE Conference, 5th International Conference on System Modeling & Advancement in Research Trends, 25th_27’h November, 2016. [23] E. Petac, A.O. Petac., “About Security Solutions in Fog Computing”, Ovidius University Press, 2016. [24] Y. Xiao, C. Zhu, “Vehicular Fog Computing: Vision and Challenges”, 2017 IEEE International Conference on Pervasive Computing and Communications. [25] X.-Q. Pham and E.-N. Huh, “Towards task scheduling in a cloud-fog computing system,” in 2016 18th AsiaPacificNetwork Operations and Management Symposium (APNOMS), Oct 2016, pp. 1–4. [26] C. Y. Huang and K. Xu, “Reliable realtime streaming in vehicular cloudfog computing networks,” in 2016 IEEE/CIC International Conference on Communications in China (ICCC), July 2016, pp. 1–6. [27] Y. Lin and H. Shen, “Cloud fog: Towards high quality of experience in cloud gaming,” in 2015 44th International Conference on Parallel Processing (ICPP), Sept 2015, pp. 500–509. [28] A. Mtibaa, K. Harras, and H. Alnuweiri, “Friend or foe? detecting and isolating malicious nodes in mobile edge computing platforms,” in 2015 IEEE 7th International Conference on Cloud Computing Technology and Science (CloudCom), Nov 2015, pp. 42–49. [29] Y. W. Law, M. Palaniswami, G. Kounga, and A. Lo, “Wake: Key management scheme for wide-area measurement systems in smart grid,” IEEE Communications Magazine, vol. 51, no. 1, pp. 34–41, 2013. [30] Z. M. Fadlullah, M. M. Fouda, N. Kato, A. Takeuchi, N. Iwasaki, and Y. Nozaki, “Toward intelligent machine-to-machine communications in smart grid,” IEEE Communications Magazine, vol. 49, no. 4, pp. 60–65, 2011. [31] F. Xu, C. C. Tan, Q. Li, G. Yan, and J. Wu. Designing a practical access point association protocol. In INFOCOM. IEEE, 2010. [32] K. Hong, D. Lillethun, U. Ramachandran, B. Ottenw¨alder, and B. Koldehofe. Opportunistic spatio-temporal event processing for mobile situation awareness. In DEBS. ACM, 2013. [33] K. Hong, D. Lillethun, U. Ramachandran, B. Ottenw¨alder, and B. Koldehofe. Mobile fog: A programming model for large-scale applications on the internet of things. In ACM SIGCOMM workshop on Mobile cloud computing, 2013. [34] S. Yi, C. Li, Q. Li, “A Survey of Fog Computing: Concepts, Applications and Issues”, MobiHoc Mobile and Ad Hoc Networking and Computing, ACM New York, NY, USA ©2015. [35] S.J. Stolfo, M.B. Salem, A.D. Keromytis, “Fog Computing: Mitigating Insider Data Theft Attacks in the Cloud”, IEEE Symposium on Security and Privacy Workshops, 2012. [36] E. Baccarelli, P.G.V. Naranjo, M.S. Scarpiniti, M. Shojafar, J.H. Abawajy, “Fog of Everything: energyefficient networked computing architectures, research challenges, and a case study”, IEEE Computer Society, 2016. [37] M. Chiang, T. Zhang, “Fog and IoT: An Overview of Research Opportunities”, IEEE Internet of Things Journal, Vol. 3, No. 6, December2016

Leave a Reply

Your email address will not be published.