IEEE Consumer Electronics Magazine - July 2017 - 76

Fog computing pushes applications,
services, data, computing power,
and decision making away from the
centralized nodes to the logical
extremes of a network.
Fog computing can be distinguished from cloud computing
based on various metrics [5]. The proximity of the fog to end
users is one of the main characteristics that differentiates the
fog from the cloud; that is, the fog resides at the edge of the
network, whereas the cloud is located within the Internet. The
cloud has a centralized geographical distribution, whereas the
fog can have a localized or distributed geographical distribution. Cloud-computing systems typically consist of only a few
resourceful server nodes, whereas the fog comprises a large
number of relatively less resourceful fog nodes. Furthermore,
processing in the fog nodes frees up core network bandwidth,
which helps to improve overall network efficiency. The distance between client and server nodes in the cloud is typically
multiple hops, whereas clients can connect to fog nodes usually through a single hop. Consequently, fog computing reduces
the latency of data transmission from IoT devices to the
offloaded server because of the proximity of the fog to end
devices as compared to cloud. Cloud-computing platforms
typically engender higher delay jitter for applications as compared to the applications running on fog nodes. Hence, fog
computing is more suitable for real-time IoT/CPS applications
than cloud computing.
The fog's ability to provide location-based customization of
content, services, and applications to IoT devices is another distinguishing characteristic of the fog. The cloud, on the other
hand, in most cases is unable to deliver specialized content, services, and applications to devices. The location-based customization of services and information is imperative because
the information may be relevant in a local context (i.e., proximity of specific geographic coordinates) and may be irrelevant
beyond the physical proximity to that location. Finally, the
cloud provides limited mobility support to end devices, whereas
mobility of end devices is better supported in the fog.
Although fog- and cloud-computing paradigms have clear
distinctions, these paradigms are not a replacement for one
another. In fact, the fog and the cloud are interdependent and
mutually beneficial, since certain functions are naturally more
advantageous to carry out in the fog, while others are better
suited to the cloud. The segmentation of what tasks go to the
fog and what tasks go to the back-end cloud is application
specific and can change dynamically based upon the state of
the network, including processor loads, link bandwidths,
storage capacities, fault events, and security threats [6]. The
cloud provides various services, such as infrastructure as a
service (IaaS), platform as a service (PaaS), and software as
a service, for organizations that require elastic scale. Fog
76 IEEE ConsumEr ElECtronICs magazInE

JULY 2017

computing can provide fog as a service (FaaS) to address various business challenges. FaaS may provide services, such
as network acceleration, network functions virtualization
(NFV), software-defined networking (SDN), content delivery, device management, complex event processing, video
encoding, protocol bridging, traffic offloading, cryptography,
and analytics platforms [6].

DISTINCTION BETWEEN FOG AND EDGE COMPUTING
The distinction between fog and edge computing is subtle.
Most of the literature has treated fog and edge computing as
synonymous and has used the terms fog and edge computing
interchangeably. We clarify the similarities and differences
between fog and edge computing here. The term mobile-edge
computing (MEC) is also often used as jargon. MEC is an
instance of edge computing where the objective is to provide
cloud-computing capabilities at the edge of the cellular network. The edge server in MEC is located at the cellular base
station. Both fog and edge computing push applications, data,
services, and computing power away from centralized nodes to
the logical extremes of a network. However, the fog-computing
paradigm has a more decentralized and distributed control compared to the edge-computing paradigm, which has relatively
more centralized control.
Another distinction between edge and fog computing is the
fog's openness, which is critical for the success of a ubiquitous
fog-computing ecosystem for IoT platforms and applications.
Proprietary or single-vendor solutions, as pursued typically in
edge computing, can engender limited supplier diversity,
which can have a negative impact on system cost, quality, market adoption, and innovation. Furthermore, the radio-access
network in an edge-computing paradigm is typically a cellular
network, whereas in fog computing, the radio-access network
can be a wireless local area network (WLAN), worldwide
interoperability for microwave access (WiMax), and/or cellular
and is in part considered a part of the fog.

RELATED WORK
Fog computing has been the subject of much research. Various
fog-computing architectures have been proposed [7]-[9], each
of which addresses specific mobility, resource management,
and optimization issues; however, a universally accepted fogcomputing architecture and standard has yet to be adopted.
Patel et al. [7] discuss the key market drivers, benefits,
requirements, objectives, and challenges of MEC. Their article also presents a high-level architectural MEC blueprint.
Aazam and Huh [8] propose a layered architecture for fog
computing to address resource-management challenges, such
as resource prediction, allocation, and pricing, in fog servers.
The authors used a probability-based model that considered
the type, traits, and characteristics of fog customers to make
these resource management decisions. Bittencourt et al. [9]
propose a layered architecture to facilitate mobility of connected IoT nodes. In their approach, a virtual machine (VM)
instance was created for each IoT node connected to the fog
server. When an IoT device crossed the radio boundary of the

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - July 2017