FOG Computing Technology

FOG Computing

Internet of Things (IoT)

• Connects internet devices – “things” (tablets, sensors, gateways, mobile-phones) to enable new forms of communication between things and people and between the IoTs themselves

• These connections create a network of IoTs

• This poses new challenges to the ways things communicate with each other and with people and the way data are manipulated once they are generated at the edges of the network

• How, what data are transmitted over the network?

• When, where data are processed or stored?

What is It? [Cisco 2015]

• Ιntroduced by Cisco, a bridge between IoTs and the Cloud

• FOG extends cloud to be closer to things that produce data

• Analyze IoT data closer to where its collected  so as to minimize latency and processing load on cloud

• Any device with computing, storage, network connectivity can be a FOG node

• Can be deployed anywhere (Factory floor, vehicle, human body, etc)

Cloud and IoT: Fog can develop anywhere in-between the two

Why FOG, What it means to Business?

• IoT speeds up awareness and response to events

• By the time data makes its way to the cloud for analysis, the opportunity to act might be gone

• Faster responses can improve output, service quality, safety

• To days cloud models are not always designed for the volume, variety and speed of data

• Moving all data to the cloud for analysis would slow down processing, responses, takes bandwidth and is expensive

More issues

• Despite Cloud’s advantages, health care, businesses, government, military organizations and entities that manage sensitive or classified data are reluctant to adopt cloud based solutions due to security risks of transferring data over the Internet

• Certain functions are naturally more advantageous to carry out in Fog while others are better suited to cloud

• Software To Data Approach: bring software to the data rather than transferring data to the cloud

• Transfer analysis results to the cloud

• FOG addresses this problem  as well

What you get ?

• FOG filters and analyzes the most time intensive data at the network edge close to where it is generated

• Milliseconds matter when trying to prevent manufacturing line shutdowns and make the difference between averting disaster and a cascading system failure

• Benefits include

• Greater business agility, security

• Deeper insights, improved privacy

• Lower operating cost (bandwidth, storage, processing)

• Sends data loads to cloud only for storage and further data analysis (big data analysis)

Application Areas

• Smart cities

• Collect data on city activities e.g. traffic  (change signals on surveillance of incoming traffic to prevent accidents or reduce congestion. Data could also be sent to the cloud for longer-term analytics)

• Wearable Technology

• Data from wearable sensors need to be processed locally to inform user and also communicated to the cloud

• Wellbeing

• monitor environmental conditions in house, health status, in house operations for improving the quality of living especially for elderly, disabled

• Industry 4.0

• a sensor on a critical machine sends readings associated with imminent failure

What happens in Fog/Cloud?

• Fog Nodes: micro data centers at network edge

• Like small clouds: cloudlets

• Intelligent controllers and gateways collect data from devices

• Receive feeds from IoTs using a protocol in real-time

• Run IoT apps for real-time control, context processing, data analytics

• Provide transient storage

• Send periodic data summaries to the cloud

• The cloud: public, private cloud etc.

• Receives and archives data summaries from Fog nodes

• Performs data analytics to gain business insight

• Can send new application rules to Fog nodes based on these insights, new business operation plans etc.

Cloud vs FOG [Chiang 2016]

• Fog and Cloud will co-exist and work together 

• FOG will carry-out substantial amount of storage at or near end-user rather than on large scale data center

• FOG will carry-out substantial amount of communication at or near the end-user rather than all routed through the backbone network

• FOG will carry-our substantial amount of management, control and configuration at or near the end-user rather than on large scale servers

• The decision on what functions move to Cloud or keep at FOG nodes is not always easy, depends on application

• FOG and Cloud are inder-dependent and mutually beneficial

Fog Architecture (by Cisco)

Example Architecture

Security, Privacy and Trust

• While FOG may enhance security, it presents new security challenges 

• User authentication at IoTs and gateways is an issue

• Each IoT has an IP address

• Easier to hack FOG nodes and IoTs

• Malicious users can read/replace/tamper IoTs and their readings (e.g. smart meters installed at consumers house), or use consumer information for profitIn large networks, probably many un-trustworthy users --> clients don’t trust each other, or are not willing to participate

• More issues: Distributed control in a decentralized, mobile crowd of IoTs

FOG Reference Architecture (RA)

• The means of describing and understanding the requirements of a domain where the architecture applies

• Proposed by OpenFog consortium: tech industry, research  and academic institutions (est. 2015), still incomplete …

• Fog RA should support at/near end-users

• Low latency storage 

• Computation  to avoid latency/network costs

• Management, network measurement, control, configuration 

• Allow analytics results to be securely copied to backend cloud

• Business deployment

Principles (Pillars) of FOG RA

• Security: end-to-end, node and network security

• Scalability: nodes, networks, storage and all services are scalable without disrupting system performance

• Openness: nodes info and functionality is transparent to applications, nodes can be created anywhere and be discovered / connected / used, while ensuring security/safety/privacy 

• Autonomy: no single point of failure 

• Programmability: nodes can be reprogrammed or updated 

• Reliability: high availability (uptime)

• Agility: transform data into actionable insights, quickly respond to changes 

• Hierarchy: not prerequisite, resources can be seen as a logical hierarchy based on the functional requirements of the IoT system 

FOG Hierarchy Examples

Examples 1 & 2

• Example 1: fog deployment hierarchy independent of the cloud

• E.g. the cloud can’t be used due to regulatory compliance, security and privacy reasons, unavailability of a central cloud in an area

• E.g. the cloud can’t be used due to regulatory compliance, security and privacy reasons, unavailability of a central cloud in an area

• Armed forces combat systems, drone operations, some healthcare systems, hospitals, and ATM banking systems

• Example 2: information processing in fog deployments located close to the infrastructure/process being managed.

• commercial building management, commercial solar panel monitoring, cable tv etc.

FOG Hierarchy Examples

Examples 3 & 4

• commercial building management, commercial solar panel monitoring, cable tv etc.

• Example 3: local fog for time-sensitive computation, the cloud is used for operational and business-related information processing

• Example 4: constrained environments in which the deployment of fog infrastructure may not be feasible or economical

• E.g. Agriculture, whether stations, connected cars

N-tier FOG Architecture

• Presentation, application processing, and data management functions are physically separated (3-tiers)

• Developers acquire the option of modifying or adding a specific layer, instead of reworking the entire application

• How many tiers in FOG: depends on number of sensors, type of work per sensor, latency between nodes, reliability/availability of nodes

• In each tier, each level acquires or computes information and shifts intelligence to higher levels

• Better organization of system intelligence

Intelligence in FOG

Node Management

• Manageability systems that can survive and manage fog nodes in all power states

• Produce reports on the state of fog nodes

• Automate discovery, registration and provision of end devices

• Gain full understanding of end devices (in terms of  their resources, health, operational state)

• More manageability aspects: system software and firmware updates, alerts on abnormal operation

• Manage events, start/stop, define data flowsSecurity analysis and response

Physical Node Safety

• Perform a security analysis and threat assessment in order to identify the needs the fog node

• Depends also on the location of the fog node and the degree of physical access to it

• Apply anti-tamper mechanisms to prevent physical or electronic attacks

• Measures: resistance (material), Evidence (prove the event), Detection (e.g. by Sensors), response (countermeasures e.g. clear sensitive data, shutdown or reset).

FOG Ref. Architecture Overview

Application-Node Services Layer

Application - Node Services

• May run in virtualized (containerized) environments

• Fog connector services: run at the south end, enable connections with Things, support various protocols, translate data to common data structures (e.g. JSON)

• Core services: separate the edge device from the application running in Fog node, collect data from the device and make them available to upper level services (e.g. cloud), or pass commands to lower level

• Support services: database, event broker, logging, scheduling, service registration, data clean-up etc.

• Analytic services: data filtering, averages, machine learning,

• Local decision (e.g. shutdown when temperature exceeds threshold), anomaly detection (malfunctioning device, intrusion detection)

• Application logic services 

• Integration services: allow outside fog nodes, users or applications to connect

• Transform data to desired format (e.g. JSON, XML)

• Accepts service requests (e.g. REST to prescribed addresses

• User interface: responsible for display and communication with applications and users:

• Status and operation of fog node, results of analytics processing, interface for node management and probably Web site