Applied Computer Systems

ISSN 2255-8691 (online)

ISSN 2255-8683 (print)
December 2021, vol. 26, no. 2, pp. 139–149
https://doi.org/10.2478/acss-2021-0017
https://content.sciendo.com

Fog Computing Algorithms: A Survey and Research

Opportunities
Shaifali P. Malukani1*, C. K. Bhensdadia2
1,2
Department of Computer Engineering, Dharmsinh Desai University, Nadiad, Gujarat, India

Abstract – The classic Internet of Things-Cloud Computing architectural aspects and algorithmic aspects of FC [5].
model faces challenges like high response latency, high bandwidth Recently, several studies have been conducted considering an
consumption, and high storage requirement with increasing
architectural point of view. Byers [6] surveyed practical
velocity and volume of generated data. Fog computing offers
better services to end users by bringing processing, storage, and deployment of fog computing considering system and
networking closer to them. Recently, there has been significant application design, software implementation, security,
research addressing architectural and algorithmic aspects of fog computing resource management and networking from an
computing. In the existing literature, a systematic study of architectural point of view. Nahaet at al. [7] identified use cases
architectural designs is widely conducted for various applications. that can get benefit from fog computing. They also listed the
Algorithms are seldom examined. Algorithms play a crucial role architectural requirements in the context of the fog computing
in fog computing. This survey aims to performing a comparative
study of existing algorithms. The study also presents a systematic use cases. Aazam and Huh [8] reviewed several existing
classification of the current fog computing algorithms and architectures to identify the research issues in big data-related
highlights the key challenges and research issues associated with application execution. Wanget et al. [9] studied existing
them. architectures of cloud/fog/edge for connected vehicles and
classified them into the categories of computer-aided and
Keywords – Algorithms, cloud computing, fog computing, computational enabled architectures. Alli and Alam [10]
Internet of Things (IoT), survey, taxonomy.
presented a survey on state-of-the-art fog computing
architectures, standards, tools and applications.
I. INTRODUCTION
Algorithmic aspects are seldom examined in the existing
This Internet of Things (IoT) is a world where everyday literature. Authors [5] have examined the existing architectures
objects are connected to the Internet and the data from the local and algorithms and compared them over various criteria.
environment is measured, collected, stored and processed. The However, a dedicated comprehensive study and comparative
advent of the IoT and mobile Internet has created an analysis of new fog algorithms are required in the rapidly
unparalleled flood of the enormous amount of data. growing field of fog computing. Fog algorithms are essential
Nonetheless, in most situations, resource-constrained IoT for the effective utilization of the developed fog architecture
systems are not adequate to process or store the generated data and the provision of secure and timely services to various fog
directly. The IoT, therefore, requires assistance from a powerful computing applications. The relevance of fog computing
computing paradigm like cloud computing (CC). However, the algorithms and the lack of systematic study of the algorithms
centralized service model of CC causes increased service delay are the impetus for this survey. This paper aims to review and
and bandwidth consumption as the volume of data increases. analyse recent fog algorithms. Relevant work needs to be
Additionally, the data exchange through an open network structured appropriately because of the diversity of the
increases the security threat to sensitive data [1]. To overcome literature on fog computing algorithms. This study represents a
the issues, a group of researchers [2] from Cisco introduced Fog comprehensive study and systematic classification of current
Computing (FC) in 2012. FC provides computation and storage literature from an algorithmic point of view. In contrast with
facilities at the network edge [3]. Latency-sensitive, bandwidth- [5], this survey compares and analyses the algorithms over
efficient and secure computing is supported by getting some of category-specific parameters for each of the categories
the computing and storage close to the end devices, rather than identified for the fog computing algorithms. This survey also
doing them in the cloud. highlights the research opportunities and key challenges
Rapidly growing fields of IoT, VANETs and the Internet of associated with each category of the fog algorithms.
Drones create numerous applications that need quick and real- The paper is organised as follows. Section II introduces the
time responses and thus can be benefited by fog computing [4]. concept of fog computing and associated technologies. Section
Recently, FC has attracted the significant attention of III discusses the research methodology and Section IV presents
researchers. Basic research in the area can be classified into the classification and review of the existing algorithms,

*
Corresponding author’s e-mail: [email protected]

©2021 Shaifali P. Malukani, C. K. Bhensdadia.

This is an open access article licensed under the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0),
in the manner agreed with Sciendo.
139
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

Section V discusses the open issues, and Section VI concludes required results that comply with the goals of this study, a set
the survey. of keywords has been chosen. Basic keywords for the search
process are Fog computing, algorithms, data, computation,
II. FOG COMPUTING AND ASSOCIATED TECHNOLOGIES security, application, communication, etc. Only papers written
The Fog is a distributed network that is closely associated in English and published in different scientific conference
with Cloud Computing and the Internet of Things. proceedings and journals have been included. Subjective
analysis has been conducted to select the diversified studies to
A. Fog Computing give a broader view of recent developments, current trends, and
Fog computing, a promising extension of CC, brings existing gaps in the subject area. The papers on fog
computing to the edge of the network. It helps develop low architectures are excluded from our survey efforts as they are
latency, bandwidth efficient, geographically distributed and better covered in other dedicated surveys. Based on these
privacy-aware applications. A basic three-tiered structure is criteria, 87 algorithms published in various research
comprised of the stratum of the cloud, the Fog Nodes (FNs), publications have been selected for this study. All the selected
and IoT/end users [11]. The IoT layer function includes sensing algorithms are published in the reputed databases like Springer
the local environment and transmitting the sensed data to the (20), IEEE (42), Elsevier (15), ACM (5), and other reputed
fog layer. A large number of FNs are hierarchically organised journals (5). Figure 1 shows a histogram of the total number of
and widely distributed between cloud and end devices. A research publications reviewed to attain the research goal and
resource-rich end device like vehicle, surveillance camera, or a the corresponding publication year.
computing node with advanced power like switches, gateways,
Fog Computing Algorithms
cellular base stations or core networking router can act as a
FN [1]. FNs may facilitate computation or temporary storage 30
services. The cloud layer facilitates extensive computation, 25
Number of Algorithms

analysis and permanent storage. 20

B. Cloud Computing 15
Cloud computing provides on-demand services of data 10
storage and computing power. Here, computing is performed 5
on shared resources rather than executing on local servers. It 0
provides enhanced elasticity, a large resource pool, reduced 2015 2016 2017 2018 2019 2020 2021
cost and anywhere-anytime access of updated resources to its
Year of Publication
users. It has three key business models: Infrastructure as a
Springer IEEE Elsevier ACM Others
Service (IaaS), Platform as a Service (PaaS), and Software as a
Service (SaaS). IaaS provides essential computing components Fig. 1. Histogram of the papers considered for this study.
such as servers, storage and networking as a service. PaaS
provides the infrastructure and tools on top of which developers IV. FOG COMPUTING ALGORITHMS
are expected to build applications. SaaS offers remote access to
The diversified research in the area of fog computing can be
software and its functions.
structured as shown in Fig. 2. A detailed discussion about each
C. Internet of Things category and corresponding subcategories is carried out in this
IoT refers to the giant network of the billions of physical section.
objects connected to the Internet, both gathering and A. Computation Management
exchanging data. Sensed data are transmitted over the Internet
Efficient management of computation activity carried out at
for analysis and storage. IoT applications are seen in several
fog enhances the lifetime and performance of FNs. The year-
domains, ranging from smart homes to smart cities, personal
wise distribution of algorithms considered in this review for
health care to smart hospitals, smart agriculture to smart
computation management is approximately 4 %, 6 %, 11 %,
industry, and a lot more. Automation of the task without human
27 %, 13 %, 33 %, and 6 % across the years 2015 to 2021,
intervention is possible due to IoT. IoT provides a better quality
respectively.
of life because of improved comfort, convenience and
management. Offloading
The resource-hungry activities of resource-limited end
III. RESEARCH METHODOLOGY devices are to be outsourced to FNs. This technique is often
The primary goal of this survey is to study the recent called offloading. However, benefits of offloading can be
development in fog computing algorithms, structure the obtained, if it is optimally decided when, where, what and how
diversified research in the area of fog algorithms and identify to offload [12]. Offloading algorithms are developed to decide
the research issues and opportunities in the field. To figure out whether the complete or partial task is to be offloaded (What to
the recent trends, this survey is conducted only on the research offload) and to decide the offloading destination, i.e., to cloud
published between January 2015 and May 2021. To attain the or best suitable FN (Where to offload). The algorithm can

140
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

suggest a method to upload (How to offload) or how much to collaboration of neighbouring fog data centers during heavy
offload to fog/cloud (How many tasks/ How much to offload). load to reduce the number of blocked requests without
Algorithms for When to offload decision determine whether or deteriorating the performance. Authors have not analysed the
not to offload. Algorithms for who offloading decision system performance when requests at the big data centre are
determine who will get the limited fog resources from the blocked and are offloaded to small data centres. In an integrated
competing end devices. Resource-consuming activities from a edge-fog environment, Chiti et al. [16] formulate the problem
device (D), Edge Device (ED), or Fog (F) can be offloaded to of minimization of the worst completion time and the energy
Fog (F), Edge Device (ED), or cloud (C). In the complete type consumption as a matching game with externalities and
of offloading, the whole task is offloaded to FN, in contrast, in incomplete preference lists. A post matching procedure is also
the partial type; a part of the task is offloaded. Offloading suggested to ensure stability. The scheme does not consider the
algorithms are developed for an interaction between a single mobile fog environment. Privacy and security related aspects
user and a fog node or fog node shared by multiple devices. are also not discussed for the proposed method. The proposed
Table I presents the summary of the available offloading method by Liu et al. [17] minimizes energy consumption, delay,
techniques. and payment costs for mobile devices. Scalarization is used to
transform the problem into a single-objective optimization.
Interior Point Method based algorithm is applied to reduce the
accumulated error and improve the calculation accuracy. The
method presented by Hassan et al. [18] considers dynamic
partitioning of the application to speed up the mobile
computation. They use a multilayer perceptron to predict the
performance of tasks in a different environment. The schemes
proposed in [17] and [18] consider the mobile environment but
privacy and security related aspects are not discussed. Shah-
Mansouri and Wong [19] view the competition among users for
fog resources as a potential game. The aim is to improve the
Quality of Experience (QoE) by reducing computation energy
and delay. The time for Nash equilibrium may increase
exponentially with the number of IoT users. They propose a
near-optimal solution to address the issue. However, their
scheme has not considered the dynamic arrival of tasks and the
issue of joint allocation of computation and communication
resources to mobile users. Du et al. [20] reduced the maximum
system cost, comprising the delay and energy consumption,
considering fairness for all user devices. They formulated the
joint optimization of offloading decision, transmit power,
bandwidth allocation and resource allocation as a mixed-integer
nonlinear programming problem. A semi-definite relaxation
and randomization are used for the solution. Their scheme lacks
the discussion on user mobility, delay of the user request,
multiple FNs and dynamic characteristics of the wireless
medium. Jazayeri et al. [21] consider a compromise between
the energy consumption and execution time while selecting the
best destination for offloading. Their method uses the Monitor-
Fig. 2. Classification of Fog Algorithms.
Analyse-Plan-Execute loop and a hidden Markov model auto-
scaling offloading to select the offloading destination. Load
Meng et al. [13] reduce total energy consumption prediction by machine learning and modeling of renewable
considering the delay constraint to decide cloud-only, cloud- energy can enhance the performance of the presented scheme.
first, fog-only, or fog-first computation. However, their scheme Jazayeri et al. [22] consider power consumption and execution
lacks the consideration of other important QoS factors like time of the modules. Their method uses the Monitor-Analyse-
authentication, confidentiality and availability. Rahbari and Plan-Execute loop and deep reinforcement learning is used to
Nickray [14] also decide the best destination for offloading decide the best destination. Baranwal and Vidyarthi [23]
under the constraints of authentication, confidentiality and propose a game-theoretic approach for computational
availability along with integrity, capacity, speed and cost in the offloading to decide which tasks to run on local servers and
mobile IoT environment. Their scheme does not implement which to run in a fog/cloud environment. They propose a non-
fault tolerance and the incorporation of machine learning cooperative nonzero-sum game with n +1 players where n is the
algorithms can improve the performance of the scheme. Fricker number of tasks and the local server is another player to
et al. [15] presented an analytical model to study the minimize the cost and to increase the energy efficiency for the

141
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

delay-sensitive and delay-tolerant tasks. In their scheme, Taneja and Davy [30] to find a capable network node with more
inclusion of other QoS metrics and the use of heuristic and resources than the application requirement. The method lacks a
metaheuristic algorithms can yield finer decisions. An dynamic strategy to handle network connectivity and node
improved contract net protocol and beetle antennae search failure at run time. Mouradian et al. [31] use tabu search based
algorithm based task offloading scheme is proposed in fog algorithm to obtain a sub-optimal solution for the large scale. A
computing networks by Li et al. [24]. In their approach, they nearly optimal solution is obtained for the small-scale system.
consider cooperation from the FNs to complete a task and Their approach does not consider the dynamic arrival of
model the task offloading process as a bidding process. After applications. A three-step methodology of Bench-marking,
receiving the bids, the task node optimally divides the task into Evaluation, and Testing (BET) is proposed by Venticinque and
subtasks to reduce the delay and energy consumption of the task Amato [32] to satisfy the application QoS requirements.
node. Their work lacks the consideration of the trustworthiness Dynamic optimization of the application deployment is to be
of the neighbour node while offloading the task. considered. Al-Tarawneh [33] uses the non-dominated sorting
TABLE I genetic algorithm to solve the placement problem considering
OFFLOADING ALGORITHMS the application criticality and security requirements. However,
application trust requirements are not considered. Nashaat et
Ref. Entities Type Decision No. of
al. [34] consider the QoE and FN computing capability to map
devices
the requests to the appropriate FN to maximize user
[13] D-F/C Complete Where Single
satisfaction. However, their approach yet is to be evaluated for
[14] D-F Complete Where Multiple
a different type of IoT application and real-time fog
[15] F-F Complete How Single environment.
[16] ED-ED/F/C Complete Where Multiple Faticanti et al. [35] perform throughput-based partitioning of
[17] D-F/C Complete How many Multiple application followed by the orchestration of application on
[18] D-F/C Partition What Single region-based infrastructure. Partitioned application is optimally
[19] D-F/C Complete Who Multiple placed using a pseudo gradient approach. The mobility and
[20] D-F/C Complete Where Multiple security requirements of an application are not taken into
[21] D-F/C Complete Where Multiple account in this scheme. Djemai et al. [36] propose a mobility-
[22] D-F/C Complete Where Multiple aware genetic algorithm for service placement in fog
[23] D-F/C Complete Where Single architecture considering the energy efficiency of the
[24] F-F Partial Where, How-many Single infrastructure and the application QoS requirement. However,
the authors have not studied the scalability of the approach.
Baranwal et al. [37] present a modified TOPSIS based policy to
Application Placement
map the application module with high application dependent
Research work to map the application components to FNs metrics with a computing instance with high computing
via a placing algorithm is carried out in the literature. metrics. However, the proposed placement policy does not
Skarlat et al. [25] use the exact optimization approach, a incorporate a pricing policy for FNs.
greedy first-fit heuristic, and a genetic algorithm to obtain the
placement to reduce application response time, deadline Resource Allocation
violations and cost. Their scheme does not discuss the fault The algorithms are proposed to efficiently allocate the
tolerance for the mobile fog landscape. Mahmud et al. [26] use limited fog resources to end users. There has been significant
fuzzy logic-based approaches that determine the Rating of research to effectively allocate fog computing (processing and
Expectation, i.e., priority value of request and Capacity Class storage resources) and/or fog networking resources (bandwidth
Score, i.e., status of FN. They obtain that linearly optimized of communication). Fig. 3 shows the classification of RA
mapping between the placement requests to a fog instance. algorithms based on the type of resources they allocate.
However, their scheme is yet to be translated and tested in a real
environment. Two backtracking algorithms and two heuristics
are proposed by Xia et al. [27] to reduce the average response
time and to improve the quality of the placement policy and
scalability. Consideration of intrinsic volatility of FN and the
migration cost is required. The latency-aware approach
proposed by Mahmud et al. [28] optimizes the number of active
Fig. 3. Resource allocation algorithms.
nodes by relocating the modules. The proposed scheme lacks
real-world implementation and customized settings by user and
Zhang et al. [38] model the Stackelberg games to set the price
mobility. Guerrero et al. [29] propose a heuristic solution to
resource and use many-to-many matching to allocate the
optimize weighted hope count for the most requested services
resources. In the offline auction-based mechanism presented by
near users. Their policy experiences degraded performance for
Jiao et al. [39], two bidding schemes – a constant demand
less requested applications in the higher workload and more
scheme and a multi-demand scheme – are proposed to optimize
service migrations. Binary tree-based searching is used by
social welfare. Multi-demand scheme gives higher bid

142
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

flexibility but gets a sub-optimal solution. Gu et al. [40] Task Scheduling

formulate computation and radio spectrum allocation problem Task scheduling techniques may schedule the task, the job,
as a MINLP problem. They model the problem as a student or the workflow. The task is defined as the smallest independent
project allocation problem. Matching is obtained to satisfy the entity which cannot be further subdivided. A job can be divided
delay and data size requirements of users. Network stability and into independent tasks. Workflow is a set of dependent tasks,
performance are improved by a user-oriented co-operation where the execution of one task depends on the execution of its
policy. Jia et al. [41] proposed a Deferred Algorithm based predecessor. Fig. 4 shows the classification of the studied task
double matching strategy for a double two-sided matching scheduling techniques.
game to maximize cost efficiency. The performance of the
scheme is to be tested considering more hierarchy levels in the
architecture and utilization of UAV (Unmanned Arial Vehicle).
Abedin et al. [42] use AHP to prioritize the requirements of
service. The formulated college admission problem is solved by
a one-to-many matching game to maximize the QoS. Later, the
best fit RA strategy is applied to ensure stability in user Fig. 4. Task scheduling algorithms.
association. Du et al. [20] propose a Bisection method for
Computation Resource Allocation to optimize the resource Workload based destination selection for the bag of tasks is
allocation. However, their scheme does not consider the impact considered by Tychalas and Karatza [51] to reduce the cost.
of user mobility, queuing delay of request, multiple FNs, Their scheme has little increase in the response time. In the
dynamic nature of the wireless environment. Yin et al. [43] work by Aladwani [52], Virtual Machines (VMs) with low,
allocate the resources to reduce the task delay using a standard medium, and high capability execute low, medium, and high
linear programming algorithm of the interior point method in priority tasks, respectively. However, their scheme does not
the smart industry environment. Their scheme does not consider consider the security aspect and location awareness in the
the limitations due to finite cloud resources and cloud selection of VMs. Zeng et al. [53] solve the problem of a
computing time. Do et al. [44] consider the problem of joint minimization of the maximum task completion time using the
resource allocation and minimizing carbon footprint in the linear programming relaxation method. Their work lacks the
content delivery network. They propose a distributed algorithm study of the memory management of the system. Yin et al. [43]
based on the proximal algorithm and alternating direction consider delay and resource requirement of the task and use
methods of multipliers. The performance of the scheme is yet resource threshold based method to select the destination in the
to be studied in the environment of multiple data centers. Ni et container based smart manufacturing environment. They show
al. [45] propose the priced time Petri net-based resource that the number of accepted tasks can be increased with a
allocation strategy. The method predicts time and price for the dynamic threshold value. The authors do not consider image
completion of the task. Resources that satisfy the task placement problems in containers and limitations due to finite
requirements are allocated to users. The scheme is to be cloud resources and cloud computing time. In the method
evaluated on different parameters like average completion time, proposed by Bitam et al. [54], the job is decomposed into tasks
fairness etc. as performance metrics. Luong et al. [46] propose and bees life algorithm is used for an optimal assignment for
a deep learning based optimal auction approach for resource each FN to minimize the CPU execution time and allocated
allocation in blockchain application. Feed-forward neural memory. Liu et al. [55] propose a scheduling algorithm for load
network is used to construct the assignment and payment balancing. Consideration of energy consumption and energy-
systems. The authors have not investigated the scenario with latency trade-off can improvise the performance of the scheme.
multiple resource providers, the dynamic arrival of devices and Pham and Huh [56] discuss a graph-based scheduling policy.
the dynamic availability of the resources. Peng et al. [47] Latency based task priority is calculated. A higher priority task
consider the multi-attribute-based auctioning method for is assigned to the processor node with maximum utility.
reasonable resource allocation using price and non-price Addition of the more constraints like the budget of the fog
attributes. Cao et al. [48] tackle the resource allocation problem provider and deadline can enhance the performance of the
using modified TwoArch2 with hierarchical clustering. Talaat system. In the method proposed by Choudhari et al. [57], tasks
et al. [49] present a resource allocation method based on are prioritized based on tolerance to the delay under deadline
reinforcement learning and genetic algorithm. Incoming constraint. If the task cannot be completed by the FN, it is
requests are distributed by observing the network traffic. Naha divided into subtasks and completed by the collaboration of the
et al. [50] propose a resource ranking algorithm that ranks remaining fog servers and cloud. The scheme lacks in
resources and resource provisioning is done in a hybrid and considering the dynamic priority. Zhao et al. [58] consider a
hierarchical manner by deadline based prioritizing the wireless fog-enabled multi-tiered content delivery network and
processing request. The algorithm does not discuss dynamic the technique for Lyapunov optimization for centralized
changes in resources and resource failures. assignment of access node at the control level. Their scheme
does not consider resource management, fading channel
statistics, traffic distributions and network dynamics. For
latency-critical applications, an optimized shortest job first

143
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

scheduling algorithm is proposed by Jamil et al. [59]. However, The most widely used objective is minimization of latency.
jobs with larger lengths might face starvation. Wang et al. [60] Latency minimization is a very broad category that includes
propose an improved genetic firework algorithm that uses the minimization of average response latency, transmission time,
explosion radius detection mechanism of fireworks and execution time, queuing time, completion time, or any
characteristics of resources and tasks. The proposed scheme combination of them. Energy minimization and cost
does not consider the energy consumption of fog devices while minimization are other popular objectives. While efficient
processing the task. Hosseinioun et al. [61] use the dynamic resource utilization, carbon footprint minimization,
voltage and frequency scaling technique for energy-aware maximization of throughput and load balancing are important
scheduling. The hybrid invasive weed optimization and culture but less popular objectives. Such objectives demand the
are used to construct a valid task sequence. In their approach attention of the researchers.
load balancing, trust and privacy are not discussed. Ghanavati
B. Data Management
et al. [62] present a dynamic fault-tolerant learning automata
task scheduling approach for optimizing response time while Fog receives a tremendous amount of data from associated
ensuring reliable execution of the tasks. However, the method IoT sensors. Data management at fog becomes important for the
does not consider the energy required by the task and available effective utilization of stored data in a timely manner. The year-
memory. wise distribution of the algorithms considered in this review for
the data management is approximately 7 %, 21 %, 14 % 29 %,
Objectives for Computation Management and 29 % across the years 2015, 2017, 2018, 2019 and 2020,
During this study, common objectives of fog algorithms for respectively.
computation management are identified. The major common Naas et al. [63] solve the data placement problem by
objectives are minimizing energy consumption, minimizing obtaining the exact solution via a single integer program. The
latency, minimizing cost, considering deadline and its combinatorial explosion was tested in the large fog
violations, load balancing, minimizing carbon footprint, infrastructure. Therefore, geo-partitioning based heuristic
considering security aspects, maximizing resource utilization, solution was proposed to reduce solving time. However, the
and maximizing the throughput. Table II and Fig. 5 show the method does not consider workload balance and data flow
objectives of the studied approaches. However, approaches may dependency in geo-partitioned subparts. Naas et al. [64] suggest
have objectives other than those listed that are already discussed a divide and conquer based approach to reduce the latency of
in detail in previous sections.
data transfer from the storage node to the consumer node using
Throughput
a k-way graph partitioning. Huang et al. [65] propose the multi
Resource Utilization replica data placement model. Their algorithm uses the pruning
Security method to choose a solution with minimum latency. Their work
involves trade between performance in reducing overall latency
Objectives

Carbon footprint
Load Balance and replica count. Multiple data placement with a budget
Deadline problem is discussed by Wang and Wu [66]. The authors try to
Cost
minimize the overall data access latency for a given total
Latency
Energy budget. Their scheme lacks in considering the different requests
with different sizes. The work in [18] uses FC for storage
0 10 20 30 40
expansion and finds optimal locations for data placement based
Publications
on frictional linear programming.
Fig. 5. Resource allocation algorithms.
Duplicate reports have redundant data and they consume
bandwidth and storage [67]. Ni et al. [68] propose a data
TABLE II deduplication scheme based on the BLS and homomorphic
OBJECTIVES OF COMPUTATION MANAGEMENT signature. In their scheme, all users who generated the data got
Publication
the reward. The chameleon hash function is used to prevent
[13], [16], [17], [19]–[24], [48], [51], [59], [61], users from claiming double rewards. Yan et al. [69] present a
Energy consumption decision tree and an interval table based scheme to determine
[62]
[13],[16]-[27],[29],[31],[34]–[37], [40]–[43], the specific server to traverse for duplicate data. Shynu et al.
Latency
[45], [46], [48],[50]–[60] [70] present convergent and Modified Elliptic Curve
[14], [17], [23], [25], [31], [35], [36], [38]–[41],
Cryptography (MECC) algorithm based data deduplication
Cost scheme to detect the redundancy at the block level.
[45], [47], [50], [51], [57], [58]
Vales et al. [71] propose an energy aware and adaptive
Deadline [25], [28]
Load Balancing [14], [47], [48], [52], [60]
distance-based data replication decision policy. The decision is
Carbon footprint [44] taken based on data popularity, distance to the consumer node
Security [14], [33] from the data node and battery of the node. Berkennou et
Resource Utilization [28], [30], [32], [34], [45] al. [72] design data popularity based on a migration and
Throughput [35], [42], [58] replication strategy over their proposed hierarchical multilayer
model to improve the response time and minimize the energy
consumption. However, their strategies are not mobility aware.

144
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

Data caching schemes facilitate caching of popular data at approximately 22 %, 11 %, 34 %, 11 % and 22 % across the
the network edge. In the scheme proposed by Al Ridhawi et years 2017 to 2021, respectively.
al. [73], frequently accessed cloud data are decomposed into Hu et al. [83] prevent unauthorized access by storing the legal
blocks and placed into fog storage. The authors propose a user’s user’s identity. Diffie-Helmen key agreement algorithm is used
historical data-based file location prediction technique. Bai et to generate session keys between fog node and user. An
al. [74] jointly optimize the fog storage hit rate and energy advanced encryption standard is used to encrypt the data. To
consumption using NSGA-II multi-objective optimization perform integrity checking, the Secure Hash Algorithm is
algorithm. implemented. Wazid et al. [84] implement a lightweight one-
Data reduction techniques reduce the amount of data way cryptographic function and bitwise XOR operation for
exchanged between the fog and the cloud to reduce bandwidth
resource constraint IoT devices to achieve secure key
consumption. Existing data reduction schemes can be majorly
management and user authentication. Elliptic curve point
classified into compression-based and prediction-based
multiplication and biometrics fuzzy extractor method is
techniques. Fu et al. [75] present compression-based data
reduction. They generate a cryptographic hash to represent the implemented for resource-rich devices and fog servers. Their
chunks of the file. The duplicate chunks are then replaced with scheme suffers from clogging attacks and it is improved by Ali
fingerprints. All the unique chunks from the file are bundled et al. [85]. Their scheme obtains the same communication cost
and compressed together. File-level deduplication is employed as the scheme in [84] but a 24 % increase in computation cost.
for compressed formats. Prediction based data reduction Lu et al. [86] employ the homomorphic paillier encryption, one-
technique proposed by Yu et al. [76] implements Dual Kalman way hash technique, and the Chinese remainder theorem for
Filter at fog and cloud with identical parameters. Hence, the aggregating IoT data and early filtering at the network edge.
same predictions at both platforms are obtained when Zuo et al. [87] suggest Attribute-Based Encryption with
simultaneously triggered. Only measured data out of the outsourced decryption as a promising solution to the chosen
prediction range are uploaded to achieve data reduction. cipher-text attack security attack. Guan et al. [88] ensure the
anonymity and authenticity of the device using a pseudonym
C. Communication Management
and pseudonym certificate. Paillier’s algorithm is used to ensure
New research challenges, including the need for new policies data privacy. A lattice-based multi-block and mixed signature
and protocols to facilitate interaction between heterogeneous scheme is constructed by Wang et al. [89] to obtain incremental
devices, have been addressed in the literature. The year-wise authentication of the updated fog data. The proposed
distribution of the algorithms considered in this review for incremental authentication scheme resists forgery attacks while
communication management is approximately 33 %, 17 %, keeping public key size and signature length in a reasonable
33 % and 17 % across the years 2017, 2019, 2020 and 2021, interval. They improve the computing speed of the algorithm
respectively. by finishing time-consuming computation using parallel
G´omez-C´ardenas et al. [77] proposed a hash-based naming computation or pre-computing. Their scheme does not prevent
strategy appropriate to the Fog-cloud environment. Guibert et the spread of the old signature while the new signature is
al. [78] incorporate Content-Centric Network for the efficiency created. The scheme proposed by Noura et al. [90] secures data
of communication and local storage. Kadhim and Seno [79] in fog computing by combining the AES-GMAC operation
present an Energy Efficient Multicast routing protocol based on mode with information dispersal over GF(2w). Their
Software Defined Networks and Fog computing for Vehicular cryptographic solution ensures data integrity, availability
networks called EEMSFV with deadline and bandwidth confidentiality and source authentication. Al-Khafajiy et
constraint. Abidoye and Kabaso [80] propose an energy- al. [91] consider the security and privacy concerns among fog
efficient hierarchical routing algorithm. They minimize the nodes during their collaboration. They identify and isolate the
total number of broadcast packets using an ant colony malicious fog node by a trust and recommendation based
optimization. However, data transmission in the network is proposed method. Their scheme lacks the consideration of the
susceptible to different attacks and needs some adequate energy consumption of the fog node during collaboration.
security solutions. Noorani and Seno [81] propose SDN and
Fog computing-based Switchable Routing to switch the E. Application Specific Algorithms
transmission between VANET infrastructure and the Internet to Lightweight algorithms for fog devices for application
select the best path for the inter-vehicle transmission. Saito et specific tasks are developed. The year-wise distribution of the
al. [82] present a Topic-based Data Transmission protocol to application-specific algorithms considered in this review is
deliver messages to target nodes. Their approach reduces the approximately 17 %, 33 %, 33 % and 17 % across the years
number of messages while the delivery ratio is smaller. 2018 to 2021, respectively.
Xu et al. [92] use fog computing to identify tumors using a
D. Algorithms for Security
modified and semi-supervised Fuzzy C Means algorithm. Work
Fog nodes might contain sensitive information and the done by Wan et al. [93] uses fog computing in the smart
personal information of the users. Therefore, algorithms to industry environment to schedule the job on the equipment by
maintain the confidentiality, integrity and availability of these employing improvised particle swarm optimization.
data have been developed. The year-wise distribution of the Vijayakumar et al. [94] used the fog for early detection of
algorithms considered in this review for the system security is mosquito based diseases. The similarity coefficient is used to

145
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

differentiate disease and fuzzy k nearest neighbour Several important aspects of fog computing have drawn the
implemented to categorize the users into infected and attention of the researchers and significant research has been
uninfected classes. Siddharth and Aghila [95] propose a random carried out in the field. The study of the existing literature is
projection and structural similarity index based light often carried out from an architectural point of view. Research
background subtraction algorithm for motion detection. Xu et on algorithmic aspects is often overlooked. This survey presents
al. [96] propose a scheme to quickly deploy emergency the comprehensive study of the research that has been
networking and communication using available devices like performed in the field so far from an algorithmic point of view.
routers and mobile devices as fog nodes in case of natural First, a taxonomy for existing fog computing algorithms is
disaster. However, the proposed scheme is yet to be tested for derived. Later, a comprehensive study and comparative analysis
diversity in the equipment and real-world environment. Ali et of the fog algorithms are carried out. Moreover, mainstream
al. [97] propose a dynamic deep hybrid Spatio-temporal neural goals for the computation management algorithms are
network (DHSTNet) to predict highly accurate traffic in every identified and less popular but important objectives are
region of a city. highlighted in order to draw researchers’ attention to them.
Finally, significant research issues for each fog algorithm
V. RESEARCH ISSUES AND OPPORTUNITIES category are identified and a discussion on future directions is
Energy efficiency. Power-conscious algorithms in all the fog carried out in this survey.
computing management aspects are required to enhance the
battery life of the devices. Dynamic workload. For the REFERENCES
heterogeneous and mobile fog environment, the dynamic [1] C. Puliafito, E. Mingozzi, F. Longo, A. Puliafito, and O. Rana,
“Fogcomputing for the internet of things: A survey,” ACM Transactions
arrival and departure of the requests are not an unpredictable on Internet Technology (TOIT), vol. 19, no. 2, pp. 1–41, Apr. 2019.
scenario. Policies to manage run time arrival of requests of https://doi.org/10.1145/3301443
offloading, service placement, resource allocation, scheduling, [2] F. Bonomi, R. Milito, J. Zhu, and S. Addepalli, “Fog computing and its
role in the internet of things,” in Proceedings of the first edition of the
or data management can improve user satisfaction. Mobility. MCC workshop on Mobile cloud computing, Aug. 2012, pp. 13–16.
Wireless sensors, vehicles and UAVs are an integral part of fog https://doi.org/10.1145/2342509.2342513
computing architecture and they observe a high amount of [3] R. Mahmud, R. Kotagiri, R.Buyya . “Fog computing: A taxonomy, survey
mobility in an unpredictable pattern. Thus, frequent migration and future directions”. in Internet of everything, Springer; 2018. p. 103–
130.
of fog services and user requests is unavoidable in a fog [4] M. Aazam, M. St-Hilaire, CH. Lung, I. Lambadaris, EN Huh." IoT
environment. Efficient handover, provision of security and resource estimation challenges and modeling in fog”, in Fog Computing
resource management after migration without service in the Internet of Things. Springer; 2018. p. 17–31.
[5] C. Mouradian, D. Naboulsi, S. Yangui, R. H. Glitho, M. J. Morrow, and
interruption are important research issues. Data dependency. P. A. Polakos, “A comprehensive survey on fog computing: Stateof-the-
Placement of the applications considering the data dependency art and research challenges,” IEEE communications surveys & tutorials,
[98] and the amount of data flow exchange among them can vol. 20, no. 1, pp. 416–464, 2017.
https://doi.org/10.1109/COMST.2017.2771153
enhance the system performance. SLA violation. Less service [6] C. C. Byers, “Architectural imperatives for fog computing: Use cases,
level violations for the allocated resources and schemes for requirements, and architectural techniques for fog-enabled IoT networks,”
compensation in case of violations are desired. Revenue and IEEE Communications Magazine, vol. 55, no. 8, pp. 14–20, Aug. 2017.
customer satisfaction. Optimum revenue for service providers https://doi.org/10.1109/MCOM.2017.1600885
[7] R. K. Naha, S. Garg, and A. Chan, “Fog computing architecture: Survey
considering customer satisfaction for the trade between fog and and challenges,” arXiv, no.1811.09047, 2018.
user is required to make a successful commercial model for fog [8] M. Aazam and E.-N. Huh, “Fog computing micro datacenter based
computing. The flexibility of the automated negotiations for the dynamic resource estimation and pricing model for IoT,” in 2015 IEEE
29th International Conference on Advanced Information Networking and
required resources and offered prices can make the fog system Applications. Gwangju, Korea (South), Mar. 2015, pp. 687–694.
more realistic. Data clusters. Performance of data positioning, https://doi.org/10.1109/AINA.2015.254
replicating and caching policies can be improved by grouping [9] H. Wang, T. Liu, B. Kim, C.-W. Lin, S. Shiraishi, J. Xie, and Z. Han,
“Architectural design alternatives based on cloud/edge/fog computing for
the data that are often accessed together and even distribution connected vehicles,” IEEE Communications Surveys & Tutorials, vol. 22,
of grouped data on fog nodes. Routing. The routing algorithms no. 4, pp. 2349–2377, Sep. 2020.
are to be developed for mobile FN with limited storage. Fault https://doi.org/10.1109/COMST.2020.3020854
[10] A. A. Alli and M. M. Alam, “The fog cloud of things: A survey on
tolerance and secure routing in the highly dynamic topology of concepts, architecture, standards, tools, and applications,” Internet of
the fog environment are considerable issues in fog architecture. Things, vol. 9, Art no. 100177, Mar. 2020.
Fog for more applications. The fog computing platform can https://doi.org/10.1016/j.iot.2020.100177
benefit a broad range of applications in the area of augmented [11] H. F. Atlam, R. J. Walters, and G. B. Wills, “Fog computing and the
internet of things: A review,” Big Data and Cognitive Computing, vol. 2,
reality, content delivery, mobile big data analytics, smart city, no. 2, p. 10, 2018. https://doi.org/10.3390/bdcc2020010
smart home, Healthcare, etc. [99]. Lightweight application [12] H. Wu, “Multi-objective decision-making for mobile cloud offloading: A
specific algorithms for all such applications are required. survey,” IEEE Access, vol. 6, pp. 3962–3976, Jan. 2018.
https://doi.org/10.1109/ACCESS.2018.2791504
[13] X. Meng, W. Wang, and Z. Zhang, “Delay-constrained hybrid
VI. CONCLUSION computation offloading with cloud and fog computing,” IEEE Access,
vol. 5, pp. 21355–21367, Sep. 2017.
The importance of fog computing has been realised with the https://doi.org/10.1109/ACCESS.2017.2748140
need for real-time bandwidth-efficient, latency-sensitive and
secure services for resource-constrained end IoT devices.

146
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

[14] D. Rahbari and M. Nickray, “Task offloading in mobile fog computing by [30] M. Taneja and A. Davy, “Resource aware placement of IoT application
classification and regression tree,” Peer-to-Peer Networking and modules in fog-cloud computing paradigm,” in 2017 IFIP/IEEE
Applications, vol. 13, pp. 1–19, Feb. 2019. Symposium on Integrated Network and Service Management (IM),
https://doi.org/10.1007/s12083-019-00721-7 Lisbon, Portugal, Vfy 2017, pp. 1222–1228.
[15] C. Fricker, F. Guillemin, P. Robert, and G. Thompson, “Analysis of an https://doi.org/10.23919/INM.2017.7987464
offloading scheme for data centers in the framework of fog computing,” [31] C. Mouradian, S. Kianpisheh, M. Abu-Lebdeh, F. Ebrahimnezhad, N. T.
ACM Transactions on Modeling and Performance Evaluation of Jahromi, and R. H. Glitho, “Application component placement in NFV-
Computing Systems (TOMPECS), vol. 1, no. 4, Art no. 16, pp. 1–18, Sep. based hybrid cloud/fog systems with mobile fog nodes,” IEEE Journal on
2016. https://doi.org/10.1145/2950047 Selected Areas in Communications, vol. 37, no. 5, pp. 1130–1143, May
[16] F. Chiti, R. Fantacci, and B. Picano, “A matching game for tasks 2019. https://doi.org/10.1109/JSAC.2019.2906790
offloading in integrated edge-fog computing systems,” Transactions on [32] S. Venticinque and A. Amato, “A methodology for deployment of IoT
Emerging Telecommunications Technologies, vol. 31, no. 2, p. Art no. application in fog,” Journal of Ambient Intelligence and Humanized
e3718, Aug. 2020. https://doi.org/10.1002/ett.3718 Computing, vol. 10, no. 5, pp. 1955–1976, 2019.
[17] L. Liu, Z. Chang, X. Guo, S. Mao, and T. Ristaniemi, “Multiobjective https://doi.org/10.1007/s12652-018-0785-4
optimization for computation offloading in fog computing,” IEEE [33] M. A. Al-Tarawneh, “Bi-objective optimization of application placement
Internet of Things Journal, vol. 5, no. 1, pp. 283–294, Dec. 2017. in fog computing environments,” Journal of Ambient Intelligence and
https://doi.org/10.1109/JIOT.2017.2780236 Humanized Computing, pp. 1–24, Feb. 2021.
[18] M. A. Hassan, M. Xiao, Q. Wei, and S. Chen, “Help your mobile https://doi.org/10.1007/s12652-021-02910-w
applications with fog computing,” in 2015 12th Annual IEEE [34] H. Nashaat, E. Ahmed, and R. Rizk, “IoT application placement algorithm
International Conference on Sensing, Communication, and Networking – based on multi-dimensional QoE prioritization model in fog computing
Workshop (SECON Workshops). Seattle, WA, USA, June 2015, pp. 1–6. environment,” IEEE Access, vol. 8, pp. 111 253–111 264, June 2020.
https://doi.org/10.1109/SECONW.2015.7328146 https://doi.org/10.1109/ACCESS.2020.3003249
[19] H. Shah-Mansouri and V. W. Wong, “Hierarchical fog-cloud computing [35] F. Faticanti, F. De Pellegrini, D. Siracusa, D. Santoro, and S. Cretti,
for IoT systems: A computation offloading game,” IEEE Internet of “Throughput-aware partitioning and placement of applications in fog
Things Journal, vol. 5, no. 4, pp. 3246–3257, Aug. 2018. computing,” IEEE Transactions on Network and Service Management,
https://doi.org/10.1109/JIOT.2018.2838022 vol. 17, no. 4, pp. 2436–2450, Dec. 2020.
[20] J. Du, L. Zhao, J. Feng, and X. Chu, “Computation offloading and https://doi.org/10.1109/TNSM.2020.3023011
resource allocation in mixed fog/cloud computing systems with min-max [36] T. Djemai, P. Stolf, T. Monteil, and J.-M. Pierson, “Mobility support for
fairness guarantee,” IEEE Transactions on Communications, vol. 66, energy and QoS aware IoT services placement in the fog,” in 2020
no. 4, pp. 1594–1608, 2018. International Conference on Software, Telecommunications and
https://doi.org/10.1109/TCOMM.2017.2787700 Computer Networks (SoftCOM), Split, Croatia, Sep. 2020, pp. 1–7.
[21] F. Jazayeri, A. Shahidinejad, and M. Ghobaei-Arani, “A latency-aware https://doi.org/10.23919/SoftCOM50211.2020.9238236
and energy-efficient computation offloading in mobile fog computing: A [37] G. Baranwal, R. Yadav, and D. P. Vidyarthi, “QoE aware IoT application
hidden Markov model-based approach,” The Journal of Supercomputing, placement in fog computing using modified-TOPSIS,” Mobile Networks
vol. 77, no. 5, pp. 4887–4916, 2021. https://doi.org/10.1007/s11227-020- and Applications, vol. 25, no. 5, pp. 1816–1832, Oct. 2020.
03476-8 https://doi.org/10.1007/s11036-020-01563-x
[22] F. Jazayeri, A. Shahidinejad, and M. Ghobaei-Arani, “Autonomous [38] H. Zhang, Y. Xiao, S. Bu, D. Niyato, F. R. Yu, and Z. Han, “Computing
computation offloading and auto-scaling the in the mobile fog computing: resource allocation in three-tier IoT fog networks: A joint optimization
a deep reinforcement learning-based approach,” Journal of Ambient approach combining stackelberg game and matching,” IEEE Internet of
Intelligence and Humanized Computing, vol. 12, pp. 8265–8284, 2021. Things Journal, vol. 4, no. 5, pp. 1204–1215, Oct. 2017.
https://doi.org/10.1007/s12652-020-02561-3 https://doi.org/10.1109/JIOT.2017.2688925
[23] G. Baranwal and D. P. Vidyarthi, “Computation offloading model for [39] Y. Jiao, P. Wang, D. Niyato, and K. Suankaewmanee, “Auction
smart factory,” Journal of Ambient Intelligence and Humanized mechanisms in cloud/fog computing resource allocation for public
Computing, vol. 12, pp. 8305–8318, 2021. blockchain networks,” IEEE Transactions on Parallel and Distributed
https://doi.org/10.1007/s12652-020-02564-0 Systems, vol. 30, no. 9, pp. 1975–1989, Sep. 2019.
[24] X. Li, Z. Zang, F. Shen, and Y. Sun, “Task offloading scheme based on https://doi.org/10.1109/TPDS.2019.2900238
improved contract net protocol and beetle antennae search algorithm in [40] Y. Gu, Z. Chang, M. Pan, L. Song, and Z. Han, “Joint radio and
fog computing networks,” Mobile Networks and Applications, vol. 25, computational resource allocation in IoT fog computing,” IEEE
no. 6, pp. 2517–2526, 2020. https://doi.org/10.1007/s11036-020-01593-5 Transactions on Vehicular Technology, vol. 67, no. 8, pp. 7475–7484,
[25] O. Skarlat, M. Nardelli, S. Schulte, M. Borkowski, and P. Leitner, Aug. 2018. https://doi.org/10.1109/TVT.2018.2820838
“Optimized IoT service placement in the fog,” Service Oriented [41] B. Jia, H. Hu, Y. Zeng, T. Xu, and Y. Yang, “Double-matching resource
Computing and Applications, vol. 11, no. 4, pp. 427–443, Oct. 2017. allocation strategy in fog computing networks based on cost efficiency,”
https://doi.org/10.1007/s11761-017-0219-8 Journal of Communications and Networks, vol. 20, no. 3, pp. 237–246,
[26] R. Mahmud, S. N. Srirama, K. Ramamohanarao, and R. Buyya, “Quality June 2018. https://doi.org/10.1109/JCN.2018.000036
of experience (QoE)-aware placement of applications in fog computing [42] S. F. Abedin, M. G. R. Alam, S. A. Kazmi, N. H. Tran, D. Niyato, and
environments,” Journal of Parallel and Distributed Computing, vol. 132, C. S. Hong, “Resource allocation for ultra-reliable and enhanced mobile
pp. 190–203, Oct. 2019. https://doi.org/10.1016/j.jpdc.2018.03.004 broadband IoT applications in fog network,” IEEE Transactions on
[27] Y. Xia, X. Etchevers, L. Letondeur, T. Coupaye, and F. Desprez, Communications, vol. 67, no. 1, pp. 489–502, Jan. 2018.
“Combining hardware nodes and software components ordering-based https://doi.org/10.1109/TCOMM.2018.2870888
heuristics for optimizing the placement of distributed IoT applications in [43] L. Yin, J. Luo, and H. Luo, “Tasks scheduling and resource allocation in
the fog,” in Proceedings of the 33rd Annual ACM Symposium on Applied fog computing based on containers for smart manufacturing,” IEEE
Computing, Apr. 2018, pp. 751–760. Transactions on Industrial Informatics, vol. 14, no. 10, pp. 4712–4721,
https://doi.org/10.1145/3167132.3167215 Oct. 2018. https://doi.org/10.1109/TII.2018.2851241
[28] R. Mahmud, K. Ramamohanarao, and R. Buyya, “Latency-aware [44] C. T. Do, N. H. Tran, C. Pham, M. G. R. Alam, J. H. Son, and C. S. Hong,
application module management for fog computing environments,” ACM “A proximal algorithm for joint resource allocation and minimizing
Transactions on Internet Technology (TOIT), vol. 19, no. 1, Art no. 9, carbon footprint in geo-distributed fog computing,” in 2015International
pp. 1–21, Mar. 2018. https://doi.org/10.1145/3186592 Conference on Information Networking (ICOIN), Cambodia, Mar. 2015,
[29] C. Guerrero, I. Lera, and C. Juiz, “A lightweight decentralized service pp. 324–329. https://doi.org/10.1109/ICOIN.2015.7057905
placement policy for performance optimization in fog computing,” [45] L. Ni, J. Zhang, C. Jiang, C. Yan, and K. Yu, “Resource allocation
Journal of Ambient Intelligence and Humanized Computing, vol. 10, strategy in fog computing based on priced timed petri nets,” IEEE Internet
no. 6, pp. 2435–2452, 2019. https://doi.org/10.1007/s12652-018-0914-0 of Things Journal, vol. 4, no. 5, pp. 1216–1228, Oct. 2017.
https://doi.org/10.1109/JIOT.2017.2709814

147
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

[46] N. C. Luong, Y. Jiao, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “A [64] M. I. Naas, L. Lemarchand, J. Boukhobza, and P. Raipin, “A graph
machine-learning-based auction for resource trading in fog computing,” partitioning-based heuristic for runtime IoT data placement strategies in a
IEEE Communications Magazine, vol. 58, no. 3, pp. 82–88, Mar. 2020. fog infrastructure,” in Proceedings of the 33rd Annual ACM Symposium
https://doi.org/10.1109/MCOM.001.1900136 on Applied Computing, Apr. 2018, pp. 767–774.
[47] X. Peng, K. Ota, and M. Dong, “Multiattribute-based double auction https://doi.org/10.1145/3167132.3167217
toward resource allocation in vehicular fog computing,” IEEE Internet of [65] T. Huang, W. Lin, Y. Li, L. He, and S. Peng, “A latency-aware multiple
Things Journal, vol. 7, no. 4, pp. 3094–3103, Apr. 2020. data replicas placement strategy for fog computing,” Journal of Signal
https://doi.org/10.1109/JIOT.2020.2965009 Processing Systems, vol. 91, no. 10, pp. 1191–1204, Feb. 2019.
[48] B. Cao, Z. Sun, J. Zhang, and Y. Gu, “Resource allocation in 5G IoV https://doi.org/10.1007/s11265-019-1444-5
architecture based on SDN and fog-cloud computing,” IEEE [66] N. Wang and J. Wu, “Latency minimization through optimal data
Transactions on Intelligent Transportation Systems, vol. 22, no. 6, placement in fog networks,” Fog Computing: Theory and Practice,
pp. 3832-3840, June 2021. https://doi.org/10.1109/TITS.2020.3048844 pp. 269–291, Apr. 2020. https://doi.org/10.1002/9781119551713.ch10
[49] F. M. Talaat, M. S. Saraya, A. I. Saleh, H. A. Ali, and S. H. Ali, “A load [67] J. Wang, “When data cleaning meets crowdsourcing,” AMPlab, UC,
balancing and optimization strategy (LBOS) using reinforcement learning Berkeley, 2015.
in fog computing environment,” Journal of Ambient Intelligence and [68] J. Ni, K. Zhang, Y. Yu, X. Lin, and X. S. Shen, “Providing task allocation
Humanized Computing, vol. 11, pp. 4951–4966, Nov. 2020. and secure deduplication for mobile crowdsensing via fog computing,”
https://doi.org/10.1007/s12652-020-01768-8 IEEE Transactions on Dependable and Secure Computing, vol. 17, no. 3,
[50] R. K. Naha, S. Garg, A. Chan, and S. K. Battula, “Deadline-based pp. 581–594, 2018. https://doi.org/10.1109/TDSC.2018.2791432
dynamic resource allocation and provisioning algorithms in Fog-Cloud [69] J. Yan, X. Wang, Q. Gan, S. Li, and D. Huang, “Secure and efficient big
environment,” Future Generation Computer Systems, vol. 104, pp. 131– data deduplication in fog computing,” Soft Computing, vol. 24, pp. 5671–
141, Mar. 2020. https://doi.org/10.1016/j.future.2019.10.018 5682, Jul. 2019. https://doi.org/10.1007/s00500-019-04215-9
[51] D. Tychalas and H. Karatza, “A scheduling algorithm for a fog computing [70] P. Shynu, R. Nadesh, V. G. Menon, P. Venu, M. Abbasi, and M. R.
system with bag-of-tasks jobs: Simulation and performance evaluation,” Khosravi, “A secure data deduplication system for integrated cloud-edge
Simulation Modelling Practice and Theory, vol. 98, Art no. 101982, Jan. networks,” Journal of Cloud Computing, vol. 9, Art no. 61, pp. 1–12, Nov.
2020. https://doi.org/10.1016/j.simpat.2019.101982 2020. https://doi.org/10.1186/s13677-020-00214-6
[52] T. Aladwani, “Scheduling IoT healthcare tasks in fog computing based on [71] R. Vales, J. Moura, and R. Marinheiro, “Energy-aware and adaptive fog
their importance,” Procedia Computer Science, vol. 163, pp. 560–569, storage mechanism with data replication ruled by spatio-temporal content
2019. https://doi.org/10.1016/j.procs.2019.12.138 popularity,” Journal of Network and Computer Applications, vol. 135,
[53] D. Zeng, L. Gu, S. Guo, Z. Cheng, and S. Yu, “Joint optimization of task pp. 84–96, June 2019. https://doi.org/10.1016/j.jnca.2019.03.001
scheduling and image placement in fog computing supported software [72] A. Berkennou, G. Belalem, and S. Limam, “A replication and migration
defined embedded system,” IEEE Transactions on Computers, vol. 65, strategy on the hierarchical architecture in the fog computing
no. 12, pp. 3702–3712, Dec. 2016. environment,” Multiagent and Grid Systems, vol. 16, no. 3, pp. 291–307,
https://doi.org/10.1109/TC.2016.2536019 Oct. 2020. https://doi.org/10.3233/MGS-200333
[54] S. Bitam, S. Zeadally, and A. Mellouk, “Fog computing job scheduling [73] I. Al Ridhawi, N. Mostafa, Y. Kotb, M. Aloqaily, and I. Abualhaol, “Data
optimization based on bees swarm,” Enterprise Information Systems, caching and selection in 5G networks using F2F communication,” in 2017
vol. 12, no. 4, pp. 373–397, 2018. IEEE 28th Annual International Symposium on Personal, Indoor, and
https://doi.org/10.1080/17517575.2017.1304579 Mobile Radio Communications (PIMRC), Montreal, QC, Canada, Oct.
[55] Z. Liu, X. Yang, Y. Yang, K. Wang, and G. Mao, “DATS: Dispersive 2017, pp. 1–6. https://doi.org/10.1109/PIMRC.2017.8292681
stable task scheduling in heterogeneous fog networks,” IEEE Internet of [74] W. Bai, H. Feng, Y. Wang, and X. Han, “Research on data cache
Things Journal, vol. 6, no. 2, pp. 3423–3436, Apr. 2018. algorithm of fog computing node,” in 2020 IEEE 11th International
https://doi.org/10.1109/JIOT.2018.2884720 Conference on Software Engineering and Service Science (ICSESS),
[56] X.-Q. Pham and E.-N. Huh, “Towards task scheduling in a cloud-fog Beijing, China, Nov. 2020, pp. 197–200.
computing system,” in 2016 18th Asia-Pacific network operations and https://doi.org/10.1109/ICSESS49938.2020.9237670
management symposium (APNOMS), Kanazawa, Japan, Nov. 2016, [75] Y. Fu, X. Qiu, and J. Wang, “F2MC: Enhancing data storage services with
pp. 1–4. https://doi.org/10.1109/APNOMS.2016.7737240 fog-toMultiCloud hybrid computing,” in 2019 IEEE 38th International
[57] T. Choudhari, M. Moh, and T.-S. Moh, “Prioritized task scheduling in Performance Computing and Communications Conference (IPCCC),
fog computing,” in Proceedings of the ACMSE’18 conference, Art no. 22, London, UK, Oct. 2019, pp. 1–6.
Mar. 2018, pp. 1–8. https://doi.org/10.1145/3190645.3190699 https://doi.org/10.1109/IPCCC47392.2019.8958748
[58] S. Zhao, Y. Yang, Z. Shao, X. Yang, H. Qian, and C.-X. Wang, “FEMOS: [76] T. Yu, X. Wang, and A. Shami, “A novel fog computing enabled temporal
Fog-enabled multitier operations scheduling in dynamic wireless data reduction scheme in IoT systems,” in GLOBECOM 2017- 2017 IEEE
networks,” IEEE Internet of Things Journal, vol. 5, no. 2, pp. 1169–1183, Global Communications Conference, Singapore, Dec. 2017, pp. 1–5.
Apr. 2018. https://doi.org/10.1109/JIOT.2018.2808280 https://doi.org/10.1109/GLOCOM.2017.8253941
[59] B. Jamil, M. Shojafar, I. Ahmed, A. Ullah, K. Munir, and H. Ijaz, “A job [77] A. Gómez-Cárdenas, X. Masip-Bruin, E. Marín-Tordera, S. Kahvazadeh,
scheduling algorithm for delay and performance optimization in fog and J. Garcia, “A hash-based naming strategy for the fog-to-cloud
computing,” Concurrency and Computation: Practice and Experience, computing paradigm,” in European Conference on Parallel Processing
vol. 32, no. 7, Art no. e5581, Apr. 2020. https://doi.org/10.1002/cpe.5581 Workshops. Lecture Notes in Computer Science. vol 10659, Springer,
[60] S. Wang, T. Zhao, and S. Pang, “Task scheduling algorithm based on Cham, pp. 316–324, 2017.
improved firework algorithm in fog computing,” IEEE Access, vol. 8, https://doi.org/10.1007/978-3-319-75178-8_26
pp. 32 385–32 394, Feb. 2020. [78] D. Guibert, J. Wu, S. He, M. Wang, and J. Li, “CC-fog: Toward content-
https://doi.org/10.1109/ACCESS.2020.2973758 centric fog networks for E-health,” in 2017 IEEE 19th International
[61] P. Hosseinioun, M. Kheirabadi, S. R. K. Tabbakh, and R. Ghaemi, “A Conference on e-Health Networking, Applications and Services
new energy-aware tasks scheduling approach in fog computing using (Healthcom), Dalian, China, Oct. 2017, pp. 1–5.
hybrid meta-heuristic algorithm,” Journal of Parallel and Distributed https://doi.org/10.1109/HealthCom.2017.8210830
Computing, vol. 143, pp. 88–96, Sep. 2020. [79] A. J. Kadhim and S. A. H. Seno, “Energy-efficient multicast routing
https://doi.org/10.1016/j.jpdc.2020.04.008 protocol based on SDN and fog computing for vehicular networks,” Ad
[62] S. Ghanavati, J. Abawajy, and D. Izadi, “Automata-based dynamic fault Hoc Networks, vol. 84, pp. 68–81, Mar. 2019.
tolerant task scheduling approach in fog computing,” IEEE Transactions https://doi.org/10.1016/j.adhoc.2018.09.018
on Emerging Topics in Computing, Oct. 2020. [80] A. P. Abidoye and B. Kabaso, “Energy-efficient hierarchical routing in
https://doi.org/10.1109/TETC.2020.3033672 wireless sensor networks based on fog computing,” EURASIP Journal on
[63] M. I. Naas, P. R. Parvedy, J. Boukhobza, and L. Lemarchand, “iFogStor: Wireless Communications and Networking, Art no. 8(2021), pp. 1–26,
an IoT data placement strategy for fog infrastructure,” in 2017 IEEE 1st Jan. 2021. https://doi.org/10.1186/s13638-020-01835-w
International Conference on Fog and Edge Computing (ICFEC), Madrid, [81] N. Noorani and S. A. H. Seno, “SDN and fog computing-based switchable
Spain, Aug. 2017, pp. 97–104. https://doi.org/10.1109/ICFEC.2017.15 routing using path stability estimation for vehicular ad hoc networks,”
Peer-to-Peer Networking and Applications, vol. 13, pp. 948–964, 2020.
https://doi.org/10.1007/s12083-019-00859-4

148
 Applied Computer Systems
_________________________________________________________________________________________________2021/26

[82] T. Saito, S. Nakamura, T. Enokido, and M. Takizawa, “Epidemic and [95] R. Siddharth and G. Aghila, “A light weight background subtraction
topic-based data transmission protocol in a mobile fog computing model,” algorithm for motion detection in fog computing,” IEEE Letters of the
in International Conference on Broadband and Wireless Computing, Computer Society, vol. 3, no. 1, pp. 17–20, 2020.
Communication and Applications. Springer, Oct. 2020, pp. 34–43. https://doi.org/10.1109/LOCS.2020.2974703
https://doi.org/10.1007/978-3-030-61108-8_4 [96] J. Xu, K. Ota, and M. Dong, “Fast deployment of emergency fog service
[83] P. Hu, H. Ning, T. Qiu, H. Song, Y. Wang, and X. Yao, “Security and for disaster response,” IEEE Network, vol. 34, no. 6, pp. 100–105, 2020.
privacy preservation scheme of face identification and resolution https://doi.org/10.1109/MNET.001.1900671
framework using fog computing in internet of things,” IEEE Internet of [97] A. Ali, Y. Zhu, and M. Zakarya, “A data aggregation based approach to
Things Journal, vol. 4, no. 5, pp. 1143–1155, Oct. 2017. exploit dynamic spatio-temporal correlations for citywide crowd flows
https://doi.org/10.1109/JIOT.2017.2659783 prediction in fog computing,” Multimedia Tools and Applications, vol. 80,
[84] M. Wazid, A. K. Das, N. Kumar, and A. V. Vasilakos, “Design of secure pp. 31401–31433, Jan. 2021.
key management and user authentication scheme for fog computing https://doi.org/10.1007/s11042-020-10486-4
services,” Future Generation Computer Systems, vol. 91, pp. 475–492, [98] F. A. Salaht, F. Desprez, and A. Lebre, “An overview of service
Feb. 2019. https://doi.org/10.1016/j.future.2018.09.017 placement problem in fog and edge computing,” ACM Computing Surveys
[85] Z. Ali, S. A. Chaudhry, K. Mahmood, S. Garg, Z. Lv, and Y. B. Zikria, (CSUR), vol. 53, no. 3, Art no. 65, pp. 1–35, June 2020.
“A clogging resistant secure authentication scheme for fog computing https://doi.org/10.1145/3391196
services,” Computer Networks, vol. 185, Art no. 107731, Feb. 2021. [99] S. Yi, Z. Hao, Z. Qin, and Q. Li, “Fog computing: Platform and
https://doi.org/10.1016/j.comnet.2020.107731 applications,” in 2015 Third IEEE workshop on hot topics in web systems
[86] R. Lu, K. Heung, A. Habibi Lashkari, and A. Ghorbani, “A lightweight and technologies (HotWeb), Washington, DC, USA, Nov. 2015, pp. 73–
privacy-preserving data aggregation scheme for fog computing-enhanced 78. https://doi.org/10.1109/HotWeb.2015.22
IoT,” IEEE Access, vol. 5, pp. 3302–3312, Mar. 2017.
https://doi.org/10.1109/ACCESS.2017.2677520 Shaifali P. Malukani has received her B. E. degree
[87] C. Zuo, J. Shao, G. Wei, M. Xie, and M. Ji, “CCA-secure ABE with from Sardar Patel University, V. V. Nagar, Gujarat,
outsourced decryption for fog computing,” Future Generation Computer India in 2011, and the M. Tech. degree from Dharmsinh
Systems, vol. 78, no. 2, pp. 730–738, Jan. 2018. Desai University, Nadiad, Gujarat, India in 2015. She is
https://doi.org/10.1016/j.future.2016.10.028 currently pursuing PhD in Computer Engineering from
[88] Z. Guan, Y. Zhang, L. Wu, J. Wu, J. Li, Y. Ma, and J. Hu, “APPA: An Dharmsinh Desai University, Nadiad, Gujarat, India.
anonymous and privacy preserving data aggregation scheme for Her areas of interest are fog computing, IoT and image
fogenhanced IoT,” Journal of Network and Computer Applications, processing.
vol. 125, pp. 82–92, Jan. 2019. https://doi.org/10.1016/j.jnca.2018.09.019 She has been working as an Assistant Professor at the
[89] F. Wang, J. Wang, and W. Yang, “Efficient incremental authentication for Department of Computer Engineering at Dharmsinh
the updated data in fog computing,” Future Generation Computer Desai University, Nadiad, Gujarat, India. She has around 10 years of teaching
Systems, vol. 114, pp. 130–137, Jan. 2021. and research experience.
https://doi.org/10.1016/j.future.2020.07.039 E-mail: [email protected]
[90] H. Noura, O. Salman, A. Chehab, and R. Couturier, “Preserving data
security in distributed fog computing,” Ad Hoc Networks, vol. 94, Art no.
C. K. Bhensdadia has received his B. E. degree from
101937, Nov. 2019. https://doi.org/10.1016/j.adhoc.2019.101937
Dharmsinh Desai University, Nadiad, Gujarat, India in
[91] M. Al-Khafajiy, T. Baker, M. Asim, Z. Guo, R. Ranjan, A. Longo,
1990, the M. Tech. degree from IIT Mumbai, India in
D. Puthal, and M. Taylor, “COMITMENT: A fog computing trust
1996 and the PhD degree from KSV, Gandhinagar,
management approach,” Journal of Parallel and Distributed Computing,
Gujarat, India in 2013. His areas of interest are fog
vol. 137, pp. 1–16, Mar. 2020. https://doi.org/10.1016/j.jpdc.2019.10.006
computing, artificial intelligence and data mining.
[92] J. Xu, H. Liu, W. Shao, and K. Deng, “Quantitative 3-D shape features
He is a Professor and Head of the Department of
based tumor identification in the fog computing architecture,” Journal of
Computer Engineering at Dharmsinh Desai
Ambient Intelligence and Humanized Computing, vol. 10, no. 8, pp. 2987–
University, Nadiad, Gujarat, India. He has around 31
2997, Feb. 2019. https://doi.org/10.1007/s12652-018-0695-5
years of teaching and research experience.
[93] J. Wan, B. Chen, S. Wang, M. Xia, D. Li, and C. Liu, “Fog computing for
He has published articles in reputed peer-reviewed national and international
energy-aware load balancing and scheduling in smart factory,” IEEE
journals and conference procedings.
Transactions on Industrial Informatics, vol. 14, no. 10, pp. 4548–4556,
E-mail: [email protected]
Oct. 2018. https://doi.org/10.1109/TII.2018.2818932
[94] V. Vijayakumar, D. Malathi, V. Subramaniyaswamy, P. Saravanan, and
R. Logesh, “Fog computing-based intelligent healthcare system for the
detection and prevention of mosquito-borne diseases,” Computers in
Human Behavior, vol. 100, pp. 275–285, Nov. 2019.
https://doi.org/10.1016/j.chb.2018.12.009

149