IEEE Systems, Man and Cybernetics Magazine - July 2022 - 23

Comparisons of characteristics between the IoT and NDN
are in Table 1.
Edge-Assisted, NDN-Based ITLC
In NDN, content routers are stationary and have abundant
storage resources, so they are able to perform in-network
caching and shorten distances among data and consumers.
This is also one of the advantages of NDN. By contrast,
IoT devices have limited caching capabilities, so they
have difficulty fulfilling in-network caching, especially in
scenarios where a large number of them are involved in
producing significant amounts of data. Moreover, the
mobility of IoT devices leads to frequent changes in locations,
so it is hard to guarantee that in-network caching
can effectively reduce distances among data and consumers.
Compared with resource-constrained IoT devices,
edge devices are fixed and have relatively abundant caching
resources [12], so they can be integrated with the NDNbased
ITLC and assume the role of content routers to help
implement in-network caching and reduce distances
among data and consumers.
Open Challenges and Possible Solutions
Although edge devices can assist in realizing in-network
caching, it remains challenging to achieve the NDN-based
ITLC, due to the different features and architectures of the
ITLC and NDN, as detailed in Table 2.
Architecture
In NDN, stationary content routers provide reverse paths,
and static servers work as providers [11], so the FIB and
reverse paths are relatively stable. These are prerequisites
for performing NDN. By contrast, in the NDN-based
ITLC, an enormous number of mobile IoT devices assume
the role of content routers and perform forwarding functions.
Meanwhile, IoT devices work as providers of data.
Consequently, the mobility of IoT devices leads to unstable
network topologies, causing a stale FIB and broken
reverse paths and ultimately resulting in the failure of
data delivery.
According to [13], network stability is mainly impacted
by the node population involved in forwarding and link performance
among mobile nodes. Hence, one possible solution
for mitigating the effect of IoT device mobility on network
stability is to construct an IoT backbone by electing forwarders
to enhance link performance and reduce the IoT
device population involved in maintaining the FIB and
reverse paths. The architecture of the edge-assisted, NDNbased
ITLC is presented in Figure 2. It contains three layers,
namely, the ITLC layer, backbone layer, and edge layer. The
ITLC layer is composed of IoT devices that cooperate to produce
data by sharing resources. The backbone layer consists
of forwarders that are elected based on metrics, such
as link durations, and are responsible for managing the FIB
and reverse paths and delivering data generated by the ITLC
layer to the edge layer for caching. The edge layer includes
edge devices with relatively abundant resources that are in
charge of performing in-network caching and maintaining
data delivered by the backbone layer.
Routing
In NDN, content routers are fixed and have powerful computing
capabilities to rapidly create and update the FIB.
Moreover, providers, such as servers, are static, and their
population is relatively steady. Based on these features,
NDN uses flooding to manage the FIB and ensure its validity.
In the ITLC, mobile IoT devices have limited processing
resources and assume the role of content routers. Because
a large number of IoT devices are involved in forwarding
and maintaining the FIB, creating and updating the FIB by
using flooding may result in broadcast storms, as demonstrated
in Figure 3(a).
In addition, ITLC members work as providers, and the
fact that IoT devices frequently join and leave the ITLC
increases the frequency of FIB updates, further deteriorating
the situation and ultimately leading to FIB staleness and
Interest
Data
Interest
Data
Figure 1. The data-centric NDN mechanism.
Table 1. The comparison of characteristics.
Model
Addresscentric
IoT
Characteristics
* Each IoT device
is assigned a
unique address.
* An address is used
to identify and
discover a specific
node.
* A source node
can acquire
data only from a
destination node,
which might not
be optimal.
Data-centric
NDN
* Neither consumers
nor providers
need addresses.
* Names are employed
to discover
optimal providers.
* Data can be
obtained from any
optimal provider.
* In-network caching
is enabled.
Results
* Address configuration
for each IoT device
incurs considerable
costs and delays, degrading
data delivery
performance.
* An address cannot be
used to seek members
because it cannot
be bound to a set of
dynamic members.
* Data cannot be
acquired from any optimal
member.
* Addressing costs and
delays are avoided.
* A name may be
bound to a group of
dynamic ITLC members.
*
ITLC members assume
the role of providers, so
data can be obtained
from any optimal ITLC
member.
* Distances among data
and consumers are
shortened.
July 2022 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE 23
IEEE Systems, Man and Cybernetics Magazine - July 2022

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - July 2022
