8 - Fog & Edge Computing

Question 1

Motivation

Answer 1

One of the biggest challenges for future trends and digital innovations like the Internet of Things (IoT), embedded artificial intelligent, or ubiquitous computing is the management, storage and processing of huge amounts of data
-> Current infrastructure will struggle to cope with the data flood

Question 2

Data Challenges

Answer 2

1. Amount of data: Rising amounts of data that have to be processed
2. Latency: High demand for low latency when providing data
3. Privacy: Especially geographical location is needed for many services
4. Inaccessability: The need for data, provided in real time, requires a good internet quality

Question 3

Fog Computing

Answer 3

Fog computing is a layered model for enabling ubiquitous access to a shared continuum of scalable computing resources. The model facilitates the deployment of distributed, latency aware applications and services , and consists of fog nodes (physical or virtual), residing between smart end devices
and centralized (cloud) services.

Question 4

Fundamentals of Fog Computing

Answer 4

fog and edge computing present new distributed architectures that help to reduce latency and support the storage, management, and processing of huge amounts of data
- fog computing architecture allows the distribution of core functions (computing, storage, communication, controlling, decision making) closer to the point where the data are generated or consumed

Question 5

Fog Nodes

Answer 5

Fog nodes are either physical components or virtual components that are tightly coupled with the smart end devices or access networks and provide computing resources to these devices.

Question 6

Typical functions of fog nodes

Answer 6

• provides some form of data management (i.e., computing or storage)
• enables communication services between network’s edge layer and the fog computing service or the centralized (cloud) computing resources
• is aware of its geographical distribution and logical location
• can be deployed in a centralized or decentralized manner

Question 7

What are the two types of fog nodes?

Answer 7

Physical: gateways, switches, routers, servers, …
Virtual: virtualized switches, virtual machines, cloudlets, …

Question 8

Fog Computings key Characteristics

Answer 8

1. Contextual Location Awareness and Low Latency
2. Geographical distribution
3. Autonomy & Heterogeneity
4. Interoperability & Federation
5. Real-Time interactions
6. Scalability and Agility of federated fog clusters

Question 9

1. Contextual Location Awareness and Low Latency

Answer 9

• A subset of the fog nodes is located close to each other, and they can easily locate every device in the vicinity
• helps to achieve the lowest possible latency because it allows fog nodes to choose the shortest communication path between fog nodes

Question 10

2. Geographical distribution

Answer 10

Distributed in environment, so that low latency is guaranteed. Keeping the data locally also avoids security
issues

Question 11

3. Autonomy & Heterogeneity

Answer 11

Autonomy: fog nodes can operate independently and make own decisions local
Heterogeneity: Fog nodes can be virtual or physical, and the functions of these nodes are also very different and can change very quickly

Question 12

4. Interoperability & Federation

Answer 12

• each fog node can provide and use services from other actors in the fog infrastructure and utilize these services to operate effectively together
• Fog computing supports hierarchical structures. Nodes can be configured to deliver the service as stand alone fog node (organized vertically) or as federated node to form clusters that provide horizontal scalability over disperse geolocations (organized horizontally)

Question 13

5. Real-Time interactions

Answer 13

real-time decision making for real-time services without any interruptions (e.g. autonomous cars)

Question 14

6. Scalability and Agility of federated fog clusters

Answer 14

• Fog computing is designed to cover millions of end devices -> can handle large scales and elastic scalability
• Each fog node merely takes care of a small part of the demand -> total pressure on cloud computing is reduced
• Fog computing is scalable because of its different clusters of fog nodes or even clusters of clusters
• Supporting elastic compute, resource pooling, data
  load changes, and network condition variations

Question 15

Predominance of wireless access

Answer 15

The large scale of wireless sensors in IoT implementations demand distributed computing power

Question 16

Support for mobility

Answer 16

Fog nodes can communicate directly with mobile devices -> enables mobile data analytics (opportunity for IoT, smart city or smart vehicles). Mobility is a key distinction between fog computing and cloud computing.

Question 17

Fog Computing Service Models

Answer 17

1. Software as a service
   - applications run on a cluster of fog nodes that are managed by the fog service provider
   - access through a thin client interface or a program interface
2. Platform as a service
   - run and deploy their applications on a platform provided by the fog service provider
   - consumer can manage the deployed applications and the application-hosting environment
3. Infrastructure as a service
   - consumer here is able to manage the operating system, storage, network, and server

Question 18

Fog Computing Deployment models (4)

Answer 18

1. Private Fog
2. Community Fog
3. Public Fog
4. Hybrid Fog

Question 19

Edge Computing

Answer 19

Edge computing refers to the enabling technologies allowing computation to be performed at the edge of the network, on downstream data on behalf of cloud services and upstream data on behalf of IoT services.

Question 20

Enabling technologies of Edge Computing (4)

Answer 20

1. Tagging Technologies (Identification)
   - Tagging technologies allow to track and count virtually any physical object
2. Sensor Technologies (Perception)
   - Sensors collect data about the real world. They can augment and complement human senses.
3. Smart Technologies (Intelligence)
   - “Things become smart” they are being equipped with data processing capabilities
4. Miniaturization Technologies (Size)
   - To embed intelligence in all kinds of physical objects, computer chips and sensors need to become ever so
   smaller

Question 21

Smart “Things” in Edge Layer

Answer 21

1. Smart Devices
   tend to be multi purpose ICT devices, operating as a single portal to access sets of popular multiple application services that may reside locally on the device or remotely on servers
2. Smart Objects
   are physical objects augmented with sensing, processing, and network capabilities that act autonomous, make sense of their situation, and interact with humans and
   other objects.
3. Smart Environment
   is a physical world that is richly and invisibly interwoven with sensors, actuators, displays, and computational
   elements, embedded seamlessly in the everyday objects of our lives, and connected through a continuous
   network.

Question 22

Challenges of Fog and Edge Computing (4)

Answer 22

1. Security
2. Heterogeneity
3. Programming Platform
4. Energy Management

Question 23

1. Security

Answer 23

• nodes are not in securely protected data centers
• fostering trust between end-devices and fog or edge nodes plays a considerable role in preserving the security and reliability of fog and edge services
• traditional authentication mechanisms using public key infrastructure cannot be applied in these cases
• intrusion detection systems are placed on the side of the fog or edge node system to detect intrusive behavior by monitoring and analyzing log files, access control policies, and user login information

Question 24

2. Heterogeneity

Answer 24

• Computational and storage capabilities of fog, edge, and cloud computing differ significantly from each other
• Variation between and within different fog and edge domains
• Data is generated from different end devices, with various processors, having to interact with each other
• Network infrastructure of fog and edge computing includes not only high speed links, but also wireless access technologies

Question 25

3. Programming Platform

Answer 25

• The computation is done in user end devices that most likely run on heterogeneous platforms
• Programming in such heterogeneous platforms is a huge challenge
• The need for a unified development framework has become indispensable

Question 26

4. Energy Management

Answer 26

• Many distributed nodes raise expected energy consumption

- Therefore, the need for new and effective energy saving protocols and architectures rises

Question 27

Use Cases of Fog & Edge Computing

Answer 27

1. Health Care
   - Cao et al. (2015) proposed FAST: A fog computing assisted distributed analytics system to monitor fall for stroke mitigation
   - They implemented fall detection algorithms and incorporated them into fog based distributed fall detection system, which distribute the analytics throughout the network by splitting the detection task between the edge devices (smart phones attached to the users) and the server (servers in the cloud)
2. Public Fogs in Buses
   - Fog provides on board video streaming, gaming and social networking services to travellers using WiFi
   - The on board Fog server connects to the Cloud through cellular networks to refresh the pre-catched contents and update application services