An IoT architecture outline

There is no generally accepted M2M systems architecture or universal

set of standards that is widely acknowledged.
State of the art today is mainly coming from a few standardization
bodies that have specified either protocols as systems components,
or system and functional architectures for various parts of a
complete end-to-end M2M architecture
For IoT Attempting to produce a single architecture consequently
results in a number of optional and conditional requirements, all
depending on the particular problem at hand or application in
focus. Nevertheless, the identified key features that are needed
when building an M2M or IoT solution can now be put together into
a larger context by proposing a single view of the main functional
capabilities.
This is not a strict and formal functional architecture, but provides a
conceptual overview. It also follows the approach of looking at the
system capabilities from a layered point of view, including
highlighting key functions that go across the layers.
Approaches that are common in describing an
architecture are
• the software approach and
• network approach that are more focused on
how functions are distributed across a
network topology.
In this architecture, the different proposed
functional layers and capabilities provided are
discussed
 Asset Layer
It not provides any functionality within a target solution, but represents the
required things for any IoT application.
The assets of interest are the real world objects and entities that are subject to
being monitored and controlled, as well as having digital representations and
identities.
Examples include vehicles and machinery, fixed infrastructures such as buildings
and utility systems, homes, and people themselves thus being inanimate as
well as animate objects.
Assets can also be of a more virtual character, being subjective representations of
parts of the real world that are of interest to a person or an organization.
Example is a set of particular routes used by trucks in a logistics use case.
Information of interest may then be traffic intensity, roadwork, or road
conditions based on the actual weather situation.
Assets are instrumented with embedded technologies that bridge the digital
realm with the physical world, and that provide the capabilities to monitor and
control the assets as well as providing identities to the assets.
 Resource Layer
The Resource Layer provides the main functional capabilities of sensing,
actuation, and embedded identities.
Sensors and actuators in various devices that may be smartphones or Wireless
Sensor Actuator Networks (WSANs), M2M devices like smart meters, or
other sensor/actuator nodes, deliver these functions.
This is also where gateways of different types are placed that can provide
aggregation or other capabilities that are closely related to these basic
resources.
• Identification of assets can be provided by different types of tags:
Ex: Radio Frequency Identification(RFID) or optical codes like bar codes or
Quick Response (QR) codes.
 Communication Layer
Communication Layer is to provide the means for connectivity between the
resources on one end and the different computing infrastructures that
host and execute service support logic and application logic on the other
end.
Different types of networks realize the connectivity
• Local Area Network (LAN) – wired(ethernet), wireless(bluetooth,zigbee)
• Wireless Personal Area Networks (WPANs; also known as Body Area
Networks, BANs) for fitness or healthcare applications,
• Home or Building Area Networks (HANs and BANs, respectively) used
in automation and control applications,
• Neighborhood Area Networks (NANs), which are used in the
Distribution Grid of a Smart Electricity Grid.
Communication can also be used in more ad hoc scenarios.
• Vehicle-to-Vehicle (V2V) is one example that can target safety applications
like collision avoidance or car platooning.
Wide Area Network (WAN)-wired, wireless, public, private
Many of the existing legacy industry specific LAN protocol stacks do not use IP
as the networking protocol, but there is a growing number of examples
where the legacy protocol stacks are migrated towards IP, for instance,
ZigBee IP, BACnet over IP, and IPv6 over Low power Wireless Personal Area
Networks (6LoWPAN) Bluetooth (IETF 6LoWPAN BTLE 2013). bridges a LAN
and a WAN, gateways are used.
From a communication layer perspective, gateways are primarily used to do
interworking or protocol translation at different levels of the protocol
stack. This can involve the physical and link layers, but it can also involve
interworking on the communications or messaging level
for example, to do interworking between a legacy protocol like ZigBee to
exchanging service operations using HTTP as the means for
communication.
 Service Support Layer
IoT applications benefit from simplification by relying on support services that
perform common and routine tasks. These support services are provided
by the Service Support Layer and are typically executing in data centers or
server farms inside organizations or in a cloud environment.
These support services can provide uniform handling of the underlying devices
and networks, thus hiding complexities in the communications and
resource layers.
Examples:
remote device management that can do remote software upgrades, remote
diagnostics or recovery, and dynamically reconfigure application processing
such as setting event filters.
Communication-related functions include selection of communication
channels if different networks can be used in parallel, for example, for
reliability purposes, and publishsubscribe and message queue
mechanisms.
Location Based Service (LBS) capabilities and various Geographic Information
System (GIS) services are also important for many IoT applications.
Of more specific relevance for IoT are services that relate to sensor
originating data and actuation services, and services that relate to
different tags like RFID.
A directory that holds information of available resources and associated
service capabilities that can function as a rendezvous mechanism is one
example. In such a directory, nodes in WSANs can publish themselves with
service descriptions and how to be reached. Applications then perform
look-ups to find which device can provide the sensor reading of interest.
Another directory service example is the Object Naming Service of EPCGlobal
(GS1 EPCGlobal 2013) that can resolve an RFID code to a URL where
information about the tagged object can be found.
Other repositories can hold information about the persistent real-world
entities that are of interest to identify, monitor, and control, such as the
Entity Directory from the SENSEI project or the Electronic Product Code
Information Services (EPCIS) of EPCGlobal.
In general, data storage for anything from raw data to knowledge
representations, and processing capabilities such as data and event
capture, filtering, and stream processing are different core common
services for many IoT applications.
 Data and Information Layer
The Data and Information Layer provides a more abstract set of functions as
its main purposes are to capture knowledge and provide advanced control
logic support.
Key concepts here include data and information models and knowledge
representation in general, and the focus is on the organization of
information.
We refer to a Knowledge Management Framework (KMF) as a collective term
to include data, information, domain-specific knowledge, actionable
services descriptions as, for example, represented by single actuators or
more complex composite sensing and actuation services, service
descriptors, rules, process or workflow descriptions, etc
The KMF needs to integrate anything from single pieces of data from
individual sensors to highly domain-specific expert knowledge into a
common knowledge fabric.
Key concepts to construct the KMF include semantic annotation, Linked Data
and building different ontologies.
Knowledge is highly dynamic, and different techniques are used to capture
knowledge as insights, as well as consume knowledge to learn, draw
conclusions, propose or even make decisions based on past experiences,
current knowledge, and predicted outcomes of certain actions.
 Application Layer
The Application Layer in turn provides the
specific IoT applications.
There is an open-ended array of different
applications
Examples include smart metering in the Smart
Grid, vehicle tracking, building automation, or
participatory sensing (PS)
 Business Layer
The Business Layer, which focuses on supporting the core business or
operations of any enterprise, organization, or individual that is interested
in IoT applications.
This is where any integration of the IoT applications into business processes
and enterprise systems takes place. The enterprise systems can, for
example, be Customer Relationship Management (CRM), Enterprise
Resource Planning (ERP), or other Business Support Systems (BSS).
The business layer also provides exposure to APIs for third parties to get
access to data and information, and can also contain support for direct
access to applications by human users;
For instance, city portal services for citizens in a smart city context, or
providing necessary data visualizations to the human workforce in a
particular enterprise.
The business layer relies on IoT applications as one set of enablers out of
many (e.g. field force automation), and takes care of necessary
orchestration and composition to support a business process workflow.
 Management, security & Data and Services

Management, as the name implies, deals with management of

various parts of the system solution related to its operation,
maintenance, administration, and provisioning.
This includes management of devices, communications
networks, and the general Information Technology (IT)
infrastructure as well as configuration and provisioning data,
performance of services delivered, etc.
Security is about protection of the system, its information and
services, from external threats or any other harm. Security
measures are usually required across all layers, for instance,
providing communication security and information security.
Trust and identity management, and authentication and
authorization, are key capabilities. From an IoT perspective,
management of privacy via, for example, anonymization, is in
many instances a specific requirement.
The final functional group of our outlined architecture is denoted
Data and Services. Data and Service processing can, from a
topological perspective, be done in a very distributed fashion
and at different levels of complexity.
Basic event filtering and simpler aggregation, such as data
averaging, can take place in individual sensor nodes in
WSANs, contextual metadata such as location and temporal
information can be added to sensor readings, and further
aggregation can take place higher up in the network topology.
More advanced processing is, for instance, data mining and
data analytics that can be done in near real-time.
This functional group thus represents the vertical flow of data
into knowledge, the abstraction of data and services in
different levels, and the process steps of extracting
knowledge.
As the knowledge layer is focused on the organization and
representation of knowledge, this functional group is focused
on the different processing steps in the data and services
value chain, thus at different levels of granularity and
abstraction.
Different technologies are used to support the different levels of
knowledge extraction, processing, reasoning, and decision-
making. Well-known technologies here include stream
processing, analytics, machine learning, reasoning, and
inferencing.
What is not reflected in the architecture outline is the
lifecycle aspect of an IoT system solution. Lifecycle
aspects of interest include the planning phase for any
deployment, and the design phase that involves both
systems integration and application development,
where APIs and SDKs are important.
The actual deployment involving the steps of
configuration and provisioning takes place before any
solution is put into actual operation.