DCS: Distributed Caching Strategy at the Edge of Vehicular Sensor Networks in Information-Centric Networking

Abstract

Information dissemination in current Vehicular Sensor Networks (VSN) depends on the physical location in which similar data is transmitted multiple times across the network. This data replication has led to several problems, among which resource consumption (memory), stretch, and communication latency due to the lake of data availability are the most crucial. Information-Centric Networking (ICN) provides an enhanced version of the internet that is capable of resolving such issues efficiently. ICN is the new internet paradigm that supports innovative communication systems with location-independent data dissemination. The emergence of ICN with VSNs can handle the massive amount of data generated from heterogeneous mobile sensors in surrounding smart environments. The ICN paradigm offers an in-network cache, which is the most effective means to reduce the number of complications of the receiver-driven content retrieval process. However, due to the non-linearity of the Quality-of-Experience (QoE) in VSN systems, efficient content management within the context of ICN is needed. For this purpose, this paper implements a new distributed caching strategy (DCS) at the edge of the network in VSN environments to reduce the number of overall data dissemination problems. The proposed DCS mechanism is studied comparatively against existing caching strategies to check its performance in terms of memory consumption, path stretch ratio, cache hit ratio, and content eviction ratio. Extensive simulation results have shown that the proposed strategy outperforms these benchmark caching strategies.

Full text

Sensors 2019, 19, 4407; doi:10.3390/s19204407 www.mdpi.com/journal/sensors
Article
DCS: Distributed Caching Strategy
at the Edge of Vehicular Sensor Networks
in Information-Centric Networking
Yahui Meng 1,†, Muhammad Ali Naeem 1,†, Rashid Ali 2, Yousaf Bin Zikria 3,*
and Sung Won Kim3,*
1 School of Science, Guangdong University of Petrochemical Technology, Maoming 525000, China;
mengyahui@gdupt.edu.cn (Y.M.), malinaeem7@gmail.com (M.A.N.)
2 School of Intelligent Mechatronics Engineering, Sejong University, Seoul 05006, Korea;
rashidali@sejong.ac.kr
3 Department of Information and Communication Engineering, Yeungnam University, Daegu 38541, Korea
* Correspondence: yousafbinzikria@ynu.ac.kr (Y.B.Z.); swon@yu.ac.kr (S.W.K.)
† Equal Authorship: Y.M. and M.A.N. contributed equally.
Received: 3 September 2019; Accepted: 9 October 2019; Published: 11 October 2019
Abstract: Information dissemination in current Vehicular Sensor Networks (VSN) depends on the
physical location in which similar data is transmitted multiple times across the network. This data
replication has led to several problems, among which resource consumption (memory), stretch, and
communication latency due to the lake of data availability are the most crucial. Information-Centric
Networking (ICN) provides an enhanced version of the internet that is capable of resolving such
issues efficiently. ICN is the new internet paradigm that supports innovative communication
systems with location-independent data dissemination. The emergence of ICN with VSNs can
handle the massive amount of data generated from heterogeneous mobile sensors in surrounding
smart environments. The ICN paradigm offers an in-network cache, which is the most effective
means to reduce the number of complications of the receiver-driven content retrieval process.
However, due to the non-linearity of the Quality-of-Experience (QoE) in VSN systems, efficient
content management within the context of ICN is needed. For this purpose, this paper implements
a new distributed caching strategy (DCS) at the edge of the network in VSN environments to reduce
the number of overall data dissemination problems. The proposed DCS mechanism is studied
comparatively against existing caching strategies to check its performance in terms of memory
consumption, path stretch ratio, cache hit ratio, and content eviction ratio. Extensive simulation
results have shown that the proposed strategy outperforms these benchmark caching strategies.
Keywords: information-centric networking (ICN); client-cache (CC); video on demand (VoD);
vehicular sensor network (VSN)
1. Introduction
The increasing demands for novel applications due to advancements in technology have led to
increased interest in finding a means by which to deliver popular data contents to remote physical
locations such as in Vehicular Sensor Networks (VSNs), mainly for Vehicular Ad Hoc Networks
(VANET) [1]. In VSNs, vehicles are equipped with diverse onboard units (sensors) for the
communication of information. The exponentially-increasing usage of the internet has posed
problems for current VSNs due to its need for diverse facilities such as the dissemination of immense
amounts of data from heterogeneous consumers along with periodic connectivity in harsh signal

Sensors 2019, 19, 4407 2 of 20
propagation, spare roadside conditions, and high levels of mobility [2,3]. It is difficult to provide
these facilities to vehicular networks using the IP-based protocols of the present, host-centric
connectivity network paradigm [2,3]. However, the Information-Centric Internet (ICN) has proposed
emerging technologies to provide novel applications to fulfill future internet requirements. In recent
years, the ICN has received significant interest from the research community because of its rapid
growth and flexible nature vis-a-vis data communication services. It delivers a unique computing
environment in which the router turns into a server. As a result, these servers can modify,
understand, and measure the surrounding environment for data dissemination [4]. The immense
growth of today’s internet traffic requires high-quality communication services because of network
congestion, which has been increasing exponentially [5]. Therefore, the internet is currently facing
several problems related to network traffic. For example, the internet requires extra content retrieval
latency along with high bandwidth consumption during data dissemination.
Moreover, the usage of resources and energy have also increased. Connected devices are
resource-constrained, and connected devices have a significant impact upon communication in
everyday life. The basic concept behind ICN technology is that all objects have to operate through
processing, identifying, and caching abilities to communicate within a diverse environment and
achieve good data dissemination performance [6]. The reason is that the current internet supports an
outdated, location-based paradigm in which all devices need to connect through IP addresses that
indicates their location. As a result, the IP-based internet is facing several issues, such as
communication latency, data searching overhead due to high network congestion, and the
dissemination of identical contents many times from remote servers [7,8]. Moreover, the IP-based
internet architecture is insufficient to achieve better results in data communication through a large
number of devices because location-based communication needs a high amount of energy; this is a
fundamental limitation of internet architecture. New research and big data technology will deliver
an enormous amount of data that will be challenging for the current, IP-based internet architecture [8,9].
ICN-based projects were designed combining several modules, such as caching, naming,
forwarding, mobility, and security. However, caching is the first module that pays full attention to
determining the ICN from the IP-based internet architecture. It delivers many benefits during data
dissemination such as short stretch, and provides fast data delivery services [10]. ICN focuses on data
delivery without location dependency. Thus, this approach makes the ICN architecture beneficial for
the internet environment. ICN does not require IP addresses for data dissemination between sources
and consumers; rather, it uses unique names to send and retrieve data contents [11]. The cache is the
most significant feature of the ICN; it is used to store the transmitted contents near the desired
consumers. In vehicular networks, vehicles obtain their required contents from neighboring vehicles
in short time periods with small stretch. Therefore, there is no need to forward the coming interests
to remote providers, and a large number of user requests can be served locally.
In the ICN, consumers send their interests directly to the network, and the whole system is
responsible for sending the corresponding data to the appropriate consumer. A copy of the
disseminated content is cached at different locations between consumers and providers, according to
the selected caching strategy. This makes it possible to store the contents in a location which is
geographically close to the consumers [12]; therefore, it can reduce latency by caching contents near
consumers, because subsequent interests will be satisfied with the cached content. The purpose of the
implementation of the in-network cache is to enhance data transmission services and reduce the high
amount of network traffic that causes link congestion and increases bandwidth consumption [13].
Moreover, in-network caching can reduce energy and resource consumption because, in the ICN,
subsequent interests do not require traversing towards remote servers [14]. ICN caching is divided
into two categories, i.e., off-path and on-path caching. In off-path caching a particular entity named
Name Resolution System (NRS) is used to broadcast the published contents’ names with their
locations. Initially, all the consumer’s interests are transmitted to the NRS, and the NSR forwards
these interests to the appropriate data sources, as shown in Figure 1 (Off-Path Caching). In on-path
caching, consumers are directly sending their Interests to the network, and the network directly sends

Sensors 2019, 19, 4407 3 of 20
back the corresponding contents to the consumer, as illustrated in Figure 1 (On-Path caching).
Therefore, it can reduce the communication and computation overhead in data dissemination [15].
Figure 1. Caching architecture, On-path and Off-path caching.
In contrast to the ICN cache strategy, there are three exceptional features to take into account
when applying ICN cache to VSNs. First, in view of their protection and selfishness, drivers of
vehicles may play a tentative role in terms of obeying the guidelines of a cache-sharing strategy [2].
Furthermore, vehicles’ frequent and dynamic topology changes increase the unpredictability of the
cache strategy [16]. In addition, vehicles have weak computational and storage resources compared
to conventional network base stations (such as access points) and routers, and the cache redundancy
of the strategy ought to be diminished [17].
Most of the work done by researchers in this domain has not taken into account and explored
the characteristics of VSNs. A vehicle-to-infrastructure scenario cache policy in VSNs is proposed
in [18]. The authors proposed an Integer Linear Programming (ILP) definition of the issue of
optimally appropriating the contents in the VSN while thinking about the accessible storage limit and
connection ability to expand the likelihood that a vehicle will be able to retrieve the desired content.
However, due to weak wireless links and mobility, vehicles cannot directly access servers or access
points (APs). Therefore, a VSN cache strategy is needed at the edge of the network. For this purpose,
this paper implements a new distributed caching strategy (DCS) at the edge of the network in VSN
environments to reduce the number of data dissemination problems. The proposed DCS mechanism
is studied comparatively against existing caching strategies to check the performance in terms of
memory consumption, path stretch ratio, cache hit ratio, and content eviction ratio.
Section 2 provides an overview of related studies. Section 3 defines the problems that still exist
in associated studies. In Section 4, the proposed model is explained. In Section 5, the performance
evaluation of related and proposed research is done using a simulation platform. In Section 6, the
paper is concluded. Finally, Section 7 presents some future directions for Vehicular Sensor Networks.
2. Related Study
ICN is an emerging environment in which devices have the ability to respond to their
surroundings with the help of caching [19]. Data dissemination is the most fundamental phenomenon
of all internet architectures, in which the current IP addresses-based internet is supported by the old
version of the architecture for data transmission between remote locations. Therefore, data is
distributed when a consumer’s interest is received [20]. The reason for this is that
the IP-based internet architecture supports location-based data dissemination that produces serious
issues for future communication processes due to the exponential increase in the amount of data
traffic. At the same time, ICN delivers location-independent data dissemination and offers lots of

Sensors 2019, 19, 4407 4 of 20
benefits in terms of improving the overall data communication process [21]. Therefore, ICN can
reduce the critical issues of the IP-based architecture, and can fulfill future internet requirements.
2.1. Client-Cache (CC)
In Client-Cache Strategy (CC), the validity of cached contents is observed. The concept of CC is
derived from central-based caching, in which the content is cached at routers that are linked to more
routers [22]. The aim of CC is to increase the validity of a given content. The validity is measured
according to the lifespan of the cached content at intermediate routers and from the publisher. The
content is selected as valid if its lifespan of at the publisher is higher than its lifespan which has been
cached at an intermediate router.
In Figure 2 (Client-Cache scenario), various interests from Consumers A and B are sent to
retrieve the Content C1. Primarily, the lifespan of Content C1, Content C2, and Content C3 are shown
by VC6, VC4, and VC5, respectively, in Figure 2. In CC, the lifespan of the content is taken as VC,
which shows the validity of the content. Therefore, the lifespans of contents C1 and C2 are higher at
the publisher than at router R5. This indicates that contents C1 and C2 should be cached at router R5;
thus, C1 will be cached at router R5, as shown in Figure 2.
Figure 2. Client-Cache.
2.2. Flexible Popularity-Based Caching Strategy (FlexPop)
The FlexPop caching strategy compiles two mechanisms to complete its content caching
procedure [23]. Primarily, it performs a content caching procedure to cache transmitted content
alongside the data routing path. It executes a second content eviction procedure if the disseminated
content does not identify the free cache space for accommodation at the intermediate routers. FlexPop
requires the maintenance of a popularity table that helps to count the number of interests at each
router for all content names. On the basis of the received interests, the popularity of a given piece of
content is calculated in the PT using the content counter and popularity tag. Initially, the content is
stored in the PT to calculate its popularity. If the content within the PT indicates that its popularity is
equal or greater than the threshold, it forwards it to the comparison table (CT). The CT is responsible
for maintaining information about the popular content. It compares the popularity of the new content
with the popularities of the previous popular content; if the new content demonstrates more
significant demand than the other content, it is labeled as popular, and the CT is shared with the
neighboring routers. When the popularity of that content reaches a threshold, the content is
forwarded to the router that has the maximum number of outgoing interfaces to be cached. If the

Sensors 2019, 19, 4407 5 of 20
cache of the router having the maximum outgoing interfaces is overflowing, the content is
recommended for caching at the router that is associated with the second-highest number of outgoing
interfaces.
Figure 3 illustrates the content caching procedure in FlexPop. Initially, two contents, C2 and C3,
are cached at router R5. Router R5 is associated with the maximum outgoing interfaces, and only two
pieces of content can reside in its cache owing to its limited capacity. Three interests from consumers
A and B are sent to router R2 to retrieve content C2. In response to the received interests, the router
R2 becomes the provider and sends content C1 to consumers A and B. At the same time, the
popularity of content C1 is measured on the basis of the received interests for content C1. According
to FlexPop, C1 gains the highest popularity, as shown by the CT in Figure 3; therefore, it is labeled
“popular” and recommended for caching at the router with the maximum number of outgoing
interfaces (i.e., router R5). However, there is no free space at router R5 for caching content C1;
therefore, it will be cached at the router having the second-highest number of outgoing interfaces.
Thus, C1 will be cached at routers R4 and R6.
Figure 3.
Flexible
Popularity-based Caching Strategy.
2.3. Centrality-based Caching Strategy (CCS)
This content caching mechanism requires two approaches. First, it determines the betweenness
centrality node by calculating the links associated with each node. Second, it decides how to cache the
transmitted content along the data routing path [24]. In this caching mechanism, the interesting content
is forwarded to the node that has the maximum number of outgoing interfaces or the maximum
number of paths associated with it. If a node is associated with a high number of data routing paths,
it has more opportunities to cache the disseminated content [25]. Figure 4 illustrates the content
caching mechanism using centrality-based caching in which Consumers A, B, and C are associated
with routers R4, R7, and R9, respectively. These consumers sent three interests to retrieve content C1,
as that content is already published in the network by the content provider (P). As the interests for
content C1 reach router R3, the required content is obtained. Therefore, router R3 acts as a provider
and transmits content C1 to the interested consumers (i.e., A, B, and C). During the transmission of
the content, each router calculates the number of data routing paths associated with it. According to
the caching nature of the CCS, router R6 is selected as the betweenness centrality router because it
has the highest number of paths associated with it along the data delivery path between the provider
and the consumers. Hence, content C1 will be cached at R5.

Sensors 2019, 19, 4407 6 of 20
Figure 4.
Centrality-Based Caching Strategy.
3. The Problem Description
ICN provides centrality-based caching strategies in which the transmitted content is cached at a
betweenness centrality location to fulfill the requirements of subsequent interests [26]. However,
these caching strategies have been facing some critical issues due to the limited capacity of cache
storage at the betweenness centrality location.
The CC tries to improve content validity, but also introduced some problems such as content
eviction ratio and the stretch ratio between the consumer and the provider, because the content's
legality must be measured at all the routers, which takes time. According to CC, the interesting
content will be cached at a betweenness centrality router that increases the number of significant
issues which occur due to caching the transmitted content only at one router, such as memory
consumption. Moreover, it increases the path length due to the high content eviction rate between
the consumer and the provider. The reason for this is that all the interests need to be forwarded to
the primary publisher due to the limited cache capacity at the betweenness centrality location.
Another issue of CC is that if a large number of interests are received for Content C, and the validity
of C is the same at the betweenness centrality node and the server, then according to the CC, Content
C will not cache at the betweenness centrality router even, it is deemed to be popular. Therefore, all
the interests for popular content will be forwarded to the main server that maximizes the stretch, and
the cache hit ratio will automatically be decreased. The amount of cache storage is limited, and it is
difficult to accommodate all the content at the betweenness centrality router. Therefore, certain
problems arise in CCS that demonstrate the increased congestion which can occur at the centrality
position, leading to a high number of evictions within short intervals of time. The reason for this is
that if the cache of the betweenness centrality position becomes full, all the interest for content must
to be forwarded to the remote provider. In addition, it does not care about the content popularities,
which increases the caching of contents with lower popularities. Thus, the overall cache hit ratio
decreases because several interests have to be accomplished from remote providers owing to the large
accommodation of less popular content. Hence, the overall caching performance is decreased [27].
FlexPop was developed to solve important problems such as high memory consumption, high
evictions, and stretch. However, it increases content homogeneity through multiple replications of
the same content. Consequently, it retains the process of content evictions and higher resource
utilization. Moreover, there is no criterion by which to choose popular content according to time
consumption. We assumed a case where three interests were generated for content C1 in 5 s, and two
for content C2 in 1 s. According to FlexPop, C1 will be the most popular because no time distinction
is included for the selection of popular content. Consequently, the most recently-used content will
remain unpopular, which causes a low cache hit ratio that affects the efficiency of the content

Sensors 2019, 19, 4407 7 of 20
dissemination and increases the content eviction ratio. Moreover, in FlexPop, two tables, PT and CT,
must be computed for each piece of content and to identify popular content, which increases the
searching overhead during the selection of popular content, because several attempts must be made
to calculate the popularity.
Consequently, this increases the source (cache) utilization. The cache size is limited compared
to the giant volume of data being communicated. Owing to the enormous number of replications of
similar content, the hit ratio cannot retain its beneficial level to strengthen the caching performance.
Another concern is the procedure of changing the cache location based on popular content, which
increases the number of eviction-caching operations caused while searching for an empty cache space
and for content that has to be replaced.
• How could the content memory consumption be minimized with an improved cache hit ratio?
• How could we enhance the caching mechanism by selecting the centrality position by reducing
the stretch ratio?
To answer these questions, a new ICN-based caching strategy is proposed that has the ability to
reduce memory consumption with a high cache hit ratio and short stretch for subsequent interests.
In addition, it has the ability to minimize content eviction operations.
4. Proposed Distributed Caching Strategy (DCS)
In previous studies, it was observed that the ideal structure of the network could affect the
overall performance of the system. Cache management is an optimal feature of content centrism, and
many researchers have focused on diverse methods of managing disseminated content over
networks. Recently, several content caching mechanisms have been developed to increase the
efficiency of in-network caching by distributing the transmitted content according to the diverse
nature of caching approaches. However, in existing caching mechanisms, several problems related to
multiple replications of homogeneous content persist, thereby increasing memory wastage. Content
caching mechanisms must implement the optimal objectives to actualize the basic concept of the
NDN cache and overcome issues in the data dissemination process which are faced by the
aforementioned caching mechanisms [28]. Consequently, in this study, a new, flexible mechanism for
content caching has been designed to improve the overall caching performance [29]. The distributed
caching strategy works on the popularities of contents. Popularity-based caching strategies are more
efficient in terms of improving content dissemination, because these strategies only cache the popular
content that can fulfill the requirements of large numbers of consumers, as compared to offensive
content. Therefore, the level of popularity of a given piece of content has a significant influence on
the caching performance. Mostly, consumers are interested in downloading popular content, and it
is a substantial undertaking to cache popular content at the central position. The reason for this is
that most incoming interests will be forwarded through the central location. Therefore, if a popular
piece of content is cached at the central location, the communication distance will be decreased
because all the interests traversing a central position will be accomplished there. Moreover, the
central position may also be used to reduce the overall bandwidth consumption. Thus, in this
strategy, it becomes more important to cache popular content at centrality positions. This caching
strategy is divided into three sections, as shown below:
Case 1
The selection of popular content in this strategy is made by taking the sum of the received
interests for a specific content name. In the DCS caching strategy, each node is associated with a
distinctive statistic table in which information about content name, interest count, and a threshold
value is stored. Whenever user interest for particular content occurs, the interest count for a specific
piece of content name is incremented with the number of received interests to calculate the popularity
of that content. The threshold is a value that is specified to measure the popularity of the content. As
a result, if the content receives a number of interests which is equal to the threshold value, it is
recommended for classification as “popular”. In earlier popularity-based caching strategies, the
threshold is used as statically defined by the strategy algorithm, as described in MPC. However, DCS

Sensors 2019, 19, 4407 8 of 20
represents the dynamic threshold to calculate the popularity of a given piece of content. According
to DCS, the threshold will be equal to half the total number of received interests for all the contents
at a router. Algorithm 1 illustrates the mechanism of selecting popular content. According to the
proposed algorithm, if the number of received interests for a particular piece of content is greater
than half the total number of interests for all the pieces of content, that content is recommended for
classification as “popular”; otherwise, it is ignored. Figure 5 illustrates the mechanism for measuring
content popularity. Suppose that 14 searches are generated for Contents C1, C2, C3, and C4, as shown
in Figure 5a. According to DCS, Content C4 recommended for classification as “popular” because it
has surpassed the threshold value as shown in Figure 5b. Hence, Content C4 is recommended for
caching at the intermediate routers along the data delivery path between the user and the provider.
Therefore, the first caching operation for popular content will be performed at the closeness centrality
router, and secondly, a copy of these contents will also be cached at the edge nodes.
(a) (b)
Figure 5. Selection of Popular Content.
Algorithm 1: Selection of Popular Content
1 def Get_Popular_Content()
2 interestRequests=[]
3 interestCount={}
4 popularContents={}
5 unpopularContents={}
6 for interest in interestRequests:
7 if interestCount.__contains__(interest):
8 interestCount[interest]+=1
9 else
10 interestCount.Add(interest,1)
11 threshold=len(interestCount)/2
12 for interest in interestCount.keys():
13 if interestCount[interest] > threshold:
14 popularContents.Add(interest, interestCount[interest])
15 else
16 unpopularContents.Add(interest, interestCount[interest])
17 return popularContents, unpopularContents

Sensors 2019, 19, 4407 9 of 20
Case 2
Popular content cannot be cached in the same way as has been already implemented. The selected
contents will be cached in chunk form to reduce the usage of memory and congestion. The reason for
this is that the betweenness centrality router is associated with a large number of other routers, which
increases the congestion in data dissemination because all interests and contents need to be
forwarded through the betweenness centrality router. Therefore, the centrality router has fewer
chances to accommodate all popular content at the same time. Thus, the new model increases the
ability to cache the maximum quantity of popular content. In DCS, when a content is selected as
popular, it is recommended for caching at the closeness centrality router in chunks, as shown in
Figure 6 (Distributed Caching Strategy).
Figure 6.
Distributed Caching Strategy.
Moreover, in terms of chunks, the cache will be used efficiently because its availability will
increase to accommodate more content in chunk form. When content is deemed to be popular, it
cannot be forwarded to the centrality router until it receives more interest; in response to first interest,
only one chunk will be delivered to the closeness centrality and edge router. For subsequent
reactions, fragments are multiplied to be forwarded for caching at the closeness centrality router and
the edge router. This process will remain functional until the content transfers in its entirety to the
centrality router. In this way, if the content was popular, but after being popular, it does not receive
any interest, then it will not be cached at the centrality router, and the cache of the centrality router
will remain unallocated to accommodate subsequent, more popular contents. In this way, DCS
resolves the problem of centrality position and uses the cache in an inefficient manner.
Case 3
If a piece of content is deemed to be popular, it will also be forwarded to the edge router at the same
time for caching at the closeness centrality router, as shown in Figure 6 (Distributed Caching
Strategy). In this way, the path stretch between the consumer and the provider will be reduced for
subsequent interests.
Moreover, this will minimize the content retrieval latency for subsequent interests, and reduce
the path link congestion by caching the content at edge routers. Therefore, all the following interests
will be satisfied with edge routers. If the content is not found at the edge router, then the interest will
be satisfied from the closeness centrality router. Moreover, the closeness centrality router is selected
for the caching of popular content, because most intents will be accomplished from the centrality
router, thereby saving bandwidth consumption with a short stretch path.
For content eviction, the Least Recent (LRU) policy is used to make room for incoming content.
The present study proposes a new, ICN-based caching strategy to improve content retrieval latency

Sensors 2019, 19, 4407 10 of 20
by reducing the path length between consumers and the provider. Moreover, it reduces the
communication path, and network congestion enhances the bandwidth consumption within the
limited cache capacity of the network routers.
Figure 6 illustrates the content the caching mechanism in DCS. In the given scenario, Consumers
A and B send multiple interests to retrieve Content C1 from the provider. After a while, content C1
becomes popular, because it has received the maximum number of interests that are required to make
content popular. Therefore, content C1 is forwarded for caching at closeness centrality router R5.
Moreover, the popular content also caches at edge routers R5 and R6. Hence, subsequent interests
from Consumers A and B will be satisfied with edge routers R5 and R6. Consequently, Consumer C
can download Content C1 from the closeness centrality router.
5. Performance Evaluation
For the evaluation of the proposed caching strategy, a simulation platform is used, in which the
SocialCCNsim simulator is selected to evaluate the caching performance. The SocialCCNsim [30]
simulator was designed to measure caching performance because, in this simulator, all the network
routers are associated with cache storage. Cesar Bernardini [31] developed SocialCCNSim based on
SONETOR [32], which is a set of utilities that generates synthetic social network traces. These social
network traces represent the interactions of users in a social network or a regular client-server
fashion. Any caching strategy can be implemented in SocialCCNSim because it was developed
especially for ICN-based caching strategies. Two ISP-level topologies were selected to perform a fair
evaluation, i.e., Abilene and GEANT. In the final stage, the DCS evaluation was done using
simulations, where the chosen parameters were cache size, catalog size, network topology, Zipf
probability model, and simulation time. In our simulations, the Zipf probability distribution is used
as the popularity model with the 𝛼 parameter varying between 0.88 and 1.2; the cache size (which
specifies the available space in every node for temporally storing content objects) ranges from 100 to
1,000 elements (1 GB to 10 GB); and the catalog (which represents the total number of contents in a
network) is 107 lements. The performance of the proposed caching strategy is evaluated in terms of
memory consumption and the stretch ratio [31].
Moreover, performance is also comparatively evaluated in terms of network contention to
measure the cache hit ratio. The proposed caching strategy is compared to ICN centrality-based
caching strategies in which FlexPop, CC, and CCS are included. Moreover, categories of contents
(User-Generated Content and Video on Demand) are selected with different cache sizes, such as 1 GB
to 10 GB. The x-axis of simulation graphs is divided into ten equal parts, in which each part shows
the capacities of the cache storage (e,g., from 1 GB to 10 GB). Accordingly, 100 elements show 1 GB
and 1000 items 10 GB of cache size. Table 1 shows the simulation parameters. The proposed strategy
is evaluated in terms of checking the performance of the most applicable metrics, i.e., memory
consumption, path stretch ratio, and cache hit ratio [33].
Table 1. Simulation Parameters.
Parameter Description
Simulation time 24 h
Topologies Abilene and GEANT
Content Size 10MB each
Catalog Size 107 elements
Cache Size 100 to 1,000 elements
α Parameter 0.88 and 1.2
Content Categories UGC and VoD
Simulator SocialCCNSim
Social Network
Topology Facebook
Traffic Source SONETOR

Sensors 2019, 19, 4407 11 of 20
Metrics Memory Consumption, Cache Hit Ratio, Stretch Ratio, and Content
Eviction Ratio.
5.1. Memory Consumption
Memory consumption shows the amount of transmitted content that can be cached while
downloading the data path for a particular time interval [34]. Consumers can download the contents
from multiple routers. In ICN, memory consumption can be clarified as the term of capacity, which
shows the volume used by interest and data contents. It can be calculated using the following
equation:
𝑀𝑒𝑚𝑜𝑟𝑦 𝐶𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛  𝑈

𝑇

100 (1)
where 𝑈

shows the memory that is utilized by the cached content and 𝑇

presents the total memory
(cache storage) of the router along the data delivery path.
The DCS performs better than CCS, CC, and FlexPop in terms of memory consumption because
it provides the ability of chunk level caching of content, thereby decreasing the usage of memory and
congestion in path links. Moreover, it delivers the most popular content near consumers, reducing
data traffic and allowing contents to move freely across the network. FlexPop and CC deliver poor
performance in terms of memory consumption because of their caching of popular content only at a
centrality router, a process that increases the traffic congestion within the limited cache capacity. The
CCS caches all the content at the betweenness centrality position without considering the content’s
popularity, thereby maximizing memory consumption. Figures 7 and 8 show the simulation results
on memory consumption using two different topologies (Abilene and GEANT). From these figures,
it can be seen that the proposed DCS caching strategy performs much better than FlexPop, CC, and
CCS. Thus, we can conclude that DCS is better at enhancing the overall performance of ICN caching
in terms of achieving efficient memory consumption.
Figure 7.
Memory Consumption on Abilene topology with UGC.

Sensors 2019, 19, 4407 12 of 20
Figure 8. Memory Consumption on GEANT Topology with VoD
5.2. Stretch
The distance travelled by an interest for a publisher (content provider) is considered as
stretch [35,36]. It can be measured using the following equation:
Stretch =
∑


∑


(2)
where ∑𝐻𝑜𝑝  𝑡𝑟𝑎𝑣𝑒𝑙𝑒𝑑


represents the number of hops traveled by an interest from the end-user
to the content provider. ∑𝑇𝑜𝑡𝑎𝑙  𝐻𝑜𝑝
||

shows the total number of hops from the user to the content
provider, and I represent the total number of received interests for a given piece of content.
As the cache capacity is small compared to the disseminating content, less content can be
accommodated within the centrality routers. Besides, CCS caches all the content without taking their
popularity into account; thus, the most popular contents have fewer chances to be cached at the
betweenness centrality position due to the unavailability of a popularity module. Hence, overall
performance is reduced in terms of a stretch, because all the interests for the most popular contents
need to be forwarded to the remote provider, thereby increasing the path length between the
consumer and the provider.
The path length is increased for each interest and response. At the same time, the CC and
FlexPop cache provide the ability to accommodate popular contents at intermediate locations for a
specific time, that can decrease the path length between consumers and providers. The reason for this
is that most interests are satisfied with the centrality positions. However, these strategies provide the
ability to store popular contents, but due to the limited capacity of the cache at the betweenness
centrality router, CC and FlexPop cannot achieve better results in terms of stretch, because both
strategies are used to cache less popular contents due to their small thresholds. On the other hand,
DCS caches content in a chunk format, increasing the possibility of accommodating more contents.
Therefore, most incoming interests are satisfied with the centrality location. Moreover, the DCS
achieves better results in terms of reducing the path stretch because it provides the ability to store
content near consumers. Furthermore, it spreads out the cache ability to store chunk level caching of
popular content that is used to increase the space available for new popular content. Moreover, DCS
caches popular content at edge routers, thereby reducing the path stretch between consumers and
providers; therefore, the proposed caching strategy delivers much better results in terms of reducing

Sensors 2019, 19, 4407 13 of 20
the overall stretch ratio. From Figures 9 and 10, results indicating that DCS performs better than CCS,
CC, and FlexPop are clearly shown.
Figure 9.
Stretch Ratio on Abilene Topology with UGC.
Figure 10.
Stretch Ratio on GEANT Topology with VoD.
5.3. Cache Hit Ratio
Cache Hit Ratio refers to the quantity of the current content hits as interests are sent [37–39] by
the consumer to the provider. It can be measured as using the following equation:
𝐶𝑎𝑐ℎ𝑒 𝐻𝑖𝑡 𝑅𝑎𝑡𝑖𝑜 
∑
ℎ𝑖𝑡



∑
󰇛ℎ𝑖𝑡



𝑚𝑖𝑠𝑠

󰇜 (3)

Sensors 2019, 19, 4407 14 of 20
Figure 11.
Cache Hit Ratio on Abilene Topology with UGC.
Figure 12.
Cache Hit Ratio on GEANT Topology with VoD.
Figures 11 and 12 show the effects of the cache hit ratio on the Abilene and GEANT topologies
using different content popularity models. Among the given figures, the DCS caching strategy
performed better in terms of a cache hit ratio with both content topologies, because DCS tries to
improve the cache allocation of popular contents. Moreover, DCS caches the most popular content at
the edge router and closeness centrality routers. Therefore, subsequent interests are satisfied from
edge routers, rather than from the remote router.
If an interest cannot be served by the edge router, it is satisfied with the closeness centrality
router. Meanwhile, the CCS approach does not define any criteria by which to handle popular content
when the cache of the centrality router is full. Therefore, all interests needed to be forwarded to the
main data source (or remote router), which increases the path length and decreases the cache hit ratio.

Sensors 2019, 19, 4407 15 of 20
In comparison to the CCS approach, the CC and FlexPop approaches performed better. However,
both strategies produce a low hit ratio, because fewer contents are accommodated between the
centrality routers. On the other hand, DCS caches the content in chunks to increase the availability of
storage space at the centrality router. Consequently, we conclude that the proposed DCS strategy
performed much better by caching content close to consumers at the network edge.
5.4. Eviction Ratio
Content eviction is also one of the significant metrics by which to measure the performance of
the caching-based ICN architecture. It can be defined as when the cache of a network node becomes
saturated and there is a need to delete some content to accommodate the newly-arriving content. It
can be calculated using the following equation:
𝐸𝑣𝑖𝑐𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑖𝑜  𝑒𝑣𝑖𝑐𝑡𝑒𝑑 𝑐𝑜𝑛𝑡𝑒𝑛𝑡
𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑛𝑡𝑒𝑛𝑡 (4)
The last number of content evictions disturbs the network throughput and reduces the cache hit
and stretch ratios. The reason for this is that all the incoming interests must be forwarded to the
distant source to download the appropriate content due to an excessive number of evictions of
popular content. Figures 13 and 14 illustrate the outcomes generated by comparisons of centrality-
based caching strategies. In the given figure, we can see that the CCS shows a high content eviction
ratio, because CCS generally caches all the contents without considering their popularity, and thus,
all arriving interests must be forwarded to the remote provider.
Figure 13.
Content Eviction Ratio on Abilene Topology with UGC.

Sensors 2019, 19, 4407 16 of 20
Figure 14.
Content Eviction Ratio on GEANT Topology with VoD.
CC and FlexPop seem to show better performance in terms of the content eviction ratio, because
both strategies are used to cache popular content at centrality routers. However, due to small and
static thresholds, these caching strategies cache the least popular contents as well, causing a high
number of content evictions. On the other hand, the proposed DCS caching strategy performed better
in terms of reducing content eviction ratio as compared to CC, CCS, and FlexPop caching strategy.
The reason is that the DCS distributes and caches the content in chunks format that increases the
overall cache storage to accommodate the new contents. Besides, it uses to cache on the most popular
content at centrality routers that increase the availability of free cache to provide popular content.
Moreover, DCS caches the least popular content at the edge routers, and therefore, the subsequent
interests are accomplished from the nearest routers. Thus, DCS minimizes the content eviction ratio
by caching the least popular content at edge routers and the most popular content at centrality
routers.
6. Conclusion
The new search and big data technology will deliver a massive amount of data that will be
difficult to handle by using the current IP-based internet architecture. The reason is that the existing
internet architecture supports the addresses based data communication which will be insufficient to
fulfill the future requirements related to location-based data transmission. Similarly, the information
dissemination in the current VSNs also depend on physical location in which similar data is
transmitted several times across the network. This data replication has led to several problems in
which resource consumption (memory), stretch, and communication latency due to the lake of data
availability, are the most crucial issues. ICN provides an enhanced version of the internet that can
provide the ability to resolve such issues efficiently. ICN is a new internet paradigm that supports
innovative communication systems with location-independent data dissemination. ICN with VSN
can handle the massive amount of data generated from heterogeneous mobile sensors in surrounding
smart environments. Therefore, new ICN paradigms are emerging as a new technology to enhance
communication processes for VSNs. Moreover, it can reduce the number of difficulties in the current
internet paradigm; it provides edge routers in a VSN that can store the disseminated content for a
specific time, while taking the required memory consumption, stretch ratio, and hit ratio into account.
To improve the performance of content dissemination in an ICN-based cache of vehicles, a new
caching strategy is proposed to provide less memory consumption, a low stretch ratio, a low content
eviction ratio, and a high cache hit ratio by caching the most desired content close to consumers.

Sensors 2019, 19, 4407 17 of 20
7. Future Directions
The requirements for enhancing the VNS infrastructure are rapidly expanding, because content
generation and dissemination require more volume than the currently network capacities.
Consumers are interested in data-needed contents, rather than data source locations. The reason for
this is that the existing internet architecture supports location-based content routing, which increases
the amount of network traffic; similar contents are transferred multiple times to satisfy consumers’
needs. This redundant content routing process generates several problems, e.g., congestion, high
bandwidth usage, and resource consumption (power and energy). Consequently, these critical
problems have to be resolved by using an efficient, scalable, and reliable (secure) architecture for the
internet [40,41]. The VNS is a new promising architecture that integrates several technologies and
communication developments for the mobile internet. It provides several benefits, using
identification and tracking technologies for wireless networks.
The most significant feature of the ICN is a cache that is used to store popular contents in order
to serve user requests. In vehicular networks, vehicles can obtain their required contents from
neighboring vehicles in short time with a small stretch [42]. Therefore, there is no need to forward
incoming interests to remote providers. A large number of interests are generated for the same
content from several vehicles, and vehicles are unable to retrieve the required content directly from
the base station within partial coverage situations [43]. In this situation, the proposed caching strategy
will significantly decrease the burden on the original provider, and will provide efficient data
dissemination services [44]. Moreover, it offers distributed intelligence for smart objects (vehicles)
[43]. VNS technology delivers benefits to mobile, interconnected nodes (vehicles), such as
informatics, telecommunication, social science, and electronics. However, VSN still faces several
complications, owing in no small part to the amount of data that is produced from heterogeneous
devices (vehicles). Numerous diverse sensors are required in VSN, thereby increasing power and
resource consumption [2]. Furthermore, VSN devices transmit a tremendous amount of content that
is difficult to manage using the current IP-based internet architecture. In these situations, DCS
introduces an enhanced scheme for data transmission across the internet, and it can overcome the
current challenges of the IP-based internet and VSN [1].
The vast number of smart devices generates a significant amount of content that can be managed
efficiently by the implementation of the DCS caching strategy. DCS provides content to network
nodes, and all the nodes can store the disseminated contents during their transmission near the
consumers at the intermediate nodes. Consequently, they can fulfill the requirements of subsequent
interests in a shorter period compared to the retrieval of content from remote content providers.
Moreover, the DCS caching strategy can reduce the power and resource consumption by caching
content near users in chunk form. Thus, if a source node in the VSN is unreachable, consumers can
still retrieve their desired content from any other caching node. The integration of DCS within the
VSN can increase the reliability of the VSN architecture by deploying content near end users [45].
Author Contributions: Y.M. and M.A.N. formulated the problem statement and proposed the solution; M.A.N.
and R.A. structured the comparative study and related work to evaluate the proposed mechanism; Y.B.Z. and
S.W.K. supervised and guided throughout the project completion.
Acknowledgments: This research was supported in part by the Brain Korea 21 Plus Program (No.
22A20130012814) funded by the National Research Foundation of Korea (NRF), in part by the MSIT(Ministry of
Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2019-
2016-0-00313) supervised by the IITP(Institute for Information & communications Technology Planning &
Evaluation), and in part by Basic Science Research Program through the National Research Foundation of Korea
(NRF) funded by the Ministry of Education (2018R1D1A1A09082266).
Conflicts of Interest: The authors declare no conflict of interest

Sensors 2019, 19, 4407 18 of 20
References
1. Zhao, W.; Qin, Y.; Gao, D.; Foh, G.H.; Chao, H-C. An efficient cache strategy in information centric
networking vehicle-to-vehicle scenario. IEEE Access 2017, 5, 12657–12667.
2. Yan, Z.; Zeadally, S.; Park, Y.-J. A novel vehicular information network architecture based on named data
networking (NDN). IEEE Int. Things J. 2014, 1, 525–532.
3. Grewe, D.; Wagner, M.; Frey, H. PeRCeIVE: Proactive caching in ICN-based VANETs. In Proceedings of the 2016
IEEE Vehicular Networking Conference (VNC), Columbus, OH, USA, 8–10 December 2016; pp. 1–8.
4. Banerjee, A.; Chen, X.; Erman, J.; Gopalakrishnan, V.; Lee, S.; Merwe, J.V.D. MOCA: A lightweight mobile
cloud offloading architecture. In Proceedings of the eighth ACM international workshop on Mobility in the
evolving internet architecture, Miami, Fl, USA, 4 October 2013; pp. 92–101.
5. Ascigil, O.; Sourlas, V.; Psaras, I.; Pavlou, G. A native content discovery mechanism for the information-
centric networks. In Proceedings of the 4th ACM Conference on Information-Centric Networking, Berlin,
Germany, 26–28 September 2017; pp. 145–155.
6. Cisco, C. Cisco. Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2016–2021; Cisco
(White paper). Available online: https://www.cisco.com/c/en/us/solutions/collateral/service-
provider/visual-networking-index-vni/white-paper-c11-738429.html (accessed on 18 February 2019).
7. Conti, M. Computer communications: Present status and future challenges. Comput. Commun. 2014, 37, 1–4.
8. Zezulka, F.; Marcon, P.; Vesely, I.; Sajdl, O. Industry 4.0–An Introduction in the phenomenon. Ifac-Pap.
2016, 49, 8–12.
9. Lee, J.; Bagheri, B.; Kao, H.-A. A cyber-physical systems architecture for industry 4.0-based manufacturing
systems. Manuf. Lett. 2015, 3, 18–23.
10. Ahlgren, B.; Dannewitz, C.; Imbrenda, C.; Kutscher, D.; Ohlman, B. A survey of information-centric
networking. IEEE Commun. Mag. 2012, 50, 26–36.
11. Zhang; M; Luo, H.; Zhang, H. A survey of caching mechanisms in information-centric networking. IEEE
Commun. Surv. Tutorials 2015, 17, 1473–1499.
12. Xu, Y.; Li, Y.; Ci, S.; Lin, T.; Chen, F. Distributed caching via rewarding: An incentive caching model for
icn. In Proceedings of the GLOBECOM 2017–2017 IEEE Global Communications Conference, Singapore,
Singapore, 4–8 December 2017; pp. 1–6.
13. Naeem, M.A.; S.A, Nor. A survey of content placement strategies for content-centric networking. In
Proceedings of the AIP Conference Proceedings, Kedah, Malaysia, 11–13 April 2016; pp. 1–6.
14. Alberti, A.M.; Casaroli, M.A.F; Singh, D.; R, da Rosa Righi. Naming and name resolution in the future
internet: Introducing the NovaGenesis approach. Future Gener. Comput. Syst. 2017, 67, 163–179.
15. Araldo; A; Rossi, D.; Martignon, F. Cost-aware caching: Caching more (costly items) for less (ISPs
operational expenditures). IEEE Trans. Parallel Distrib. Syst. 2016, 27, 1316–1330.
16. Bian, C.; Zhao, T.; Li, X.; Yan, W. Boosting named data networking for efficient packet forwarding in urban
VANET scenarios. In Proceedings of the The 21st IEEE International Workshop on Local and Metropolitan
Area Networks, Beijing, China, 22–24 April 2015; pp. 1316–1330.
17. Grassi, G.; Pesavento, D.; Pau, G.; Vuyyuru, R.; Wakikawa, R.; Zhang, L. VANET via named data
networking. In Proceedings of the 2014 IEEE conference on computer communications workshops
(INFOCOM WKSHPS), Toronto, Canada, 27 April–2 May 2014; pp. 17–30.
18. Mauri, G.; Gerla, M.; Bruno, F.; Cesana, M.; Verticale, G. Optimal Content Prefetching in NDN Vehicle-to-
Infrastructure Scenario. IEEE Trans. Veh. Technol. 2017, 66, 2513–2525.
19. Gubbi, J.; Gerla, M.; Bruno, F.; Cesana, M.; Verticale, G. Internet of Things (IoT): A vision, architectural
elements, and future directions. Future Gener. Comput. Syst. 2013, 29, 1645–1660.
20. Zhang, Z.; Cho, M. C. Y.; Wang, C.; Hsu, C.; Chen, C.; Shieh, S. IoT security: Ongoing challenges and
research opportunities. In Proceedings of the IEEE 7th International Conference on Service-Oriented
Computing and Applications (SOCA), Matsue Japan, 17–19 November 2014; pp. 230–234.
21. Zhang, L.; Afanasyev, A.; Burke, J.; Jacobson, V.; claffy, K.; Crowley, P.; Papadopoulos, C.; Wang, L.; Zhang,
B. Named Data Networking. SIGCOMM Comput. Commun. Rev. 2014, 44, 66–73.

Sensors 2019, 19, 4407 19 of 20
22. Meddeb, M.; Dhraief, A.; Belghith, A.; Monteil, T.; Drira, K. How to cache in ICN-based IoT environments?
In Proceedings of the 2017 IEEE/ACS 14th International Conference on Computer Systems and
Applications (AICCSA), Hammamet, Tunisia, 30 October–3 November 2017; pp. 1117–1124.
23. Abani, N.; Braun, T.; Gerla, M. Proactive caching with mobility prediction under uncertainty in
information-centric networks. In Proceedings of the 4th ACM Conference on Information-Centric
Networking, New York, NY, USA, 26-28 September 2017; pp. 88–97.
24. Yan, H.; Gao, D.; Su, W.; Foh, C.H.; Zhang, H.; Vasilakos, A.V. Caching strategy based on hierarchical
cluster for named data networking. IEEE Access 2017, 5, 8433–8443.
25. Lal, K.N; Kumar, A. A centrality-measures based caching scheme for content-centric networking (CCN).
Multimed. Tools Appl. 2018, 77, 17625–17642.
26. Amadeo, M.; Campolo, C.; Molinaro, A. NDNe: Enhancing named data networking to support
cloudification at the edge. IEEE Commun. Lett. 2016. 20, 2264–2267.
27. Hassan, S.; Din, I.U.; Habbal, A.; Zakaria, N.H. A popularity based caching strategy for the future Internet.
In Proceedings of the 2016 ITU Kaleidoscope: ICTs for a Sustainable World (ITU WT), Bangkok, Thailand,
14–16 November 2016; pp. 68–74.
28. Dräxler, M.; H. Karl. Efficiency of on-path and off-path caching strategies in information centric networks.
In Proceedings of the 2012 IEEE International Conference on Green Computing and Communications,
Besancon, France, 20–23 November 2012; pp. 581–587.
29. Zhang, G.; Y. Li.; T. Lin. Caching in information centric networking: A survey. Comput. Netw. 2013, 57,
3128–3141.
30. Bernardini, C. Stratégies de Cache basées sur la popularité pour Content Centric Networking; PhD Thesis,
Lorraine university, University of Lorraine, Nancy, France, 2015.
31. Bernardini, C.; Silverston, T.; Festor, O. SONETOR: A social network traffic generator. In Proceedings of
the 2014 IEEE International Conference on Communications (ICC), Sydney, Australia, 10–14 June 2014;
pp. 3734–3739.
32. Bernardini, C.; Silverston, T.; Festor, O. MPC: Popularity-based caching strategy for content centric
networks. In Proceedings of the 2013 IEEE International Conference on Communications (ICC), Budapest,
Hungary, 9–13 June 2013; pp. 3619–3623.
33. Bernardini, C.; Silverston, T.; Festor, O. A comparison of caching strategies for content centric networking.
In Proceedings of the 2015 IEEE Global Communications Conference (GLOBECOM), San Diego, CA, USA,
6–10 December 2015; 1–6.
34. Sheather, S.J.; M.C. Jones. A reliable data-based bandwidth selection method for kernel density estimation.
J. R. Stat. Soc.: Ser. B (Methodol.) 1991, 53, 683–690.
35. Badenhop, C.W.; Graham, S.; Ramsey, B.W.; Mullins, B.; Mailloux, L.O. The Z-Wave routing protocol and
its security implications. Comput. Secur. 2017, 68, 112–129.
36. Dias, J.A.F.F. Performance of management solutions and cooperation approaches for vehicular delay-
tolerant networks. Available online: http://hdl.handle.net/10400.6/4501 (accessed on 2 September 2019).
37. Beckmann, N.; Chen, H.; Cidon, A. LHD: Improving Cache Hit Rate by Maximizing Hit Density. In
Proceedings of the 15th {USENIX} Symposium on Networked Systems Design and Implementation ({NSDI}
18), Renton, WA, USA, 9-11 April 2018; pp. 389–403.
38. Chen, P.; Yue, J.; Liao, X.; Jin, H. Trade-off between Hit Rate and Hit Latency for Optimizing DRAM Cache.
Ieee Trans. Emerg. Top. Comput. 2018, 41, 1–1.
39. Tseng, F.-H.; Chien, W-C.; Wang, S-J.; Lai, C.F.; Chao, H-C. A novel cache scheme based on content
popularity and user locality for future internet. In Proceedings of the 27th Wireless and Optical
Communication Conference (WOCC), Hualien, Taiwan, 30 April–1 May 2018; pp. 1–5.
40. Li, S.; Zhang, Y.; Raychaudhuri, D.; Ravindran, R. A comparative study of MobilityFirst and NDN based
ICN-IoT architectures. In Proceedings of the 10th International Conference on Heterogeneous Networking
for Quality, Reliability, Security and Robustness, Rhodes, Greece, 18–20 August 2014; pp. 189–103.
41. Bai, X.; Liu, S.; Zhang, P.; Kantola, P. ICN: Interest-based clustering network. In Proceedings of the Fourth
International Conference on Peer-to-Peer Computing, Zurich, Switzerland, 27 August 2004; pp. 489–497.
42. Modesto, F; Boukerche, A. A novel service-oriented architecture for information-centric vehicular
networks. In Proceedings of the 19th ACM International Conference on Modeling, Analysis and Simulation
of Wireless and Mobile Systems, Malta, Malta, 13–17 November 2016; pp. 239–246.

Sensors 2019, 19, 4407 20 of 20
43. Khan, J.A.; Ghamri-Doudane, Y. Saving: Socially aware vehicular information-centric networking. IEEE
Commun. Mag. 2016, 54, 100–107.
44. Rainer, B.; Petscharnig, S. Challenges and Opportunities of Named Data Networking in Vehicle-To-
Everything Communication: A Review. Inf. 2018, 9, 264.
45. Modesto, F.M.; Boukerche, A. Seven: A novel service-based architecture for information-centric vehicular
network. Comput. Commun. 2018, 117, 133–146.
© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons
Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).