INTERNET OF THINGS
ARCHITECTING SOFTWARE SOLUTIONS
By:Mayank Vijh
AGENDA
• Introduction• Trends and Hype cycle• Major IOT Players• Real World Problems• IOT Architecture and variations• Challenges and Tools• Difference between M2M and IOT practice.• KPI and Criteria• Evaluation and Decision• Conclusion
“The Internet of Things is theintelligent connectivity of physicaldevices driving massive gains inefficiency, business growth, andquality of life.”
INRODUCTIONNew Operational EfficienciesData you collect from your factory floor, logistics network, and supply chain can reduce inventory, downtime due to maintenance, and time to market. You can also use that data to simplify operations.
Improved Safety and SecuritySensors and video cameras can help monitor equipment to improve workplace safety and guard against physical threats. Connected incident response can coordinate multiple teams to resolve situations faster.
Distributed Intelligence and ControlMore frequent, remote software upgrades and enhancements can extend the efficiency and value of your resources, products, and services.
Faster and Better Decision MakingDistributing intelligence and control offloads repetitive decisions and can help prioritize decisions that people need to make.
New Business Opportunities and Revenue StreamsMore and new ways to analyze data can help identify new potential markets and business opportunities.
TRENDS AND HYPE CYCLE
MAJOR IOT PLAYERS
REAL WORLD PROBLEMS
• Real time monitoring of devices in manufacturing industries to provide valuable decisions.
• Cost of Utilities (e.g.. Electricity, Oil, Gas, Water)• Helping Health Industry - health industry costs and improve outcomes by allowing
more people to be monitored and treated in their own home rather than in hospitals.
• Security and Protecting your privacy – IOT applications can help users remotely monitor their homes and offices.
HOW IOT HELPS
BEFORE AFTER
HIGH-LEVEL IOT ARCHITECTURE
HIGH-LEVEL IOT ARCHITECTURE
VARIATIONS
1. Gateway-Centric architecture2. Smartphone-Centric architecture3.Thing-Centric architecture4. Cloud-Centric architecture5. Enterprise-Centric architecture
GATEWAY-CENTRIC ARCHITECTURE
• Add some devices as a gateway to aggregate things together
• Things only need to sense, act on command and communicate to gateway
ADVANTAGES & DISADVANTAGESAdvantages DisadvantagesReduce the cost and complexity on the edge devices One more “Tier” adds complexity in terms of integration
Gateway can act as the connector hub for things with different data standards and wireless protocols that provide an uniform face to the outside.
Resources still reside locally comparing with using cloud
Easy to monitor and control things as a whole rather than interface with each individual of them
one extra point of failure
Gateway can control what data to be sent to Internet and provide security capability than things
Convenient to link legacy equipment into the IoT
EXAMPLES
Building system, Smart city
Such as Commons in University• Apply some rules to control
• Aircons• Lightings• Window curtains.
• Monitor and control the status of them as a whole
• Collect data for further analysis
SMARTPHONE-CENTRIC ARCHITECTURE
Phone • Phone or tablet connects the thing and Internet. When they are connected, extra functionality will be provided
• The thing has some application logic and data storage capability
ADVANTAGES & DISADVANTAGESAdvantages Disadvantages
Smartphone is meant to be acting as a gateway for many IoT use cases because• Almost carried by everyone• Sufficient computing power• Internet connection• Mobility
Smartphones are not standardized, OS, constrains
Reduce the cost, complexity, need for a user interface, power consumption and weight of the edge devices
Requires a smartphone for full functionality and connectivity
User can decide whether send data to Internet or not and what to send. It may help to protect privacy.
Security of the things depends on the security of the phone
EXAMPLES
The use cases of this style include • Home-automation• Wearable like smart watch, fitbit etc
To connect with an smartphone may make them become smarter.
THING-CENTRIC ARCHITECTURE
Things are smart on their own and major computing resources are located on board. Things are self-sufficient and communicate to the Internet only for centralized coordination and analysis
THING-CENTRIC ARCHITECTURE
ADVANTAGES & DISADVANTAGESAdvantages Disadvantages Real-time functionality and response are more possible,
because resources are local and immediate. Cost and complexity are added to the thing and the edge
of the network.
It works well when connectivity is problematic or not desired.
If designed improperly, it can lead to many islands of remote things and computing resources
It makes the thing as independent as possible — not relying on the Internet for resources.
The cost and performance advantages of the cloud are not used. This also makes the centralized "big picture" view of a set of assets harder to obtain.
It allows for more security and privacy by requiring only minimal communications via the Internet.
It requires thing-specific knowledge which may be hard to find.
It saves on communications costs, since data is selectively sent out or received, while the remainder is stored on board.
EXAMPLESGlowCaps bottles reminds you to take medicines, refill bottle
and doctor coordination via a wireless chip.
Smart Belly trash can alert municipal services to know when a bin needs to be emptied. This information can drastically reduce the number of pick-ups required
HP smart Printer: HP has a new service called Instant Ink in the U.S. When you sign up for Instant Ink and connect your Wi-Fi-enabled printer to the Internet, it will monitor your ink levels, telling HP when you're running low. Before you run out of ink, a new ink cartridge magically arrives.
EXAMPLESSmart lighting system from Echelon allows a city to intelligently provide the
right level of lighting needed by time of day, season, and weather conditions. Cities have shown a reduction in street lighting energy use by up to 30% using solutions like this.
Smart Cars: • on-board diagnostic systems that alert you about almost everything in the car,
from faulty lights to tire pressure, • navigational system guiding you to destination• Cars talking to each other and to road infrastructure about road conditions, weather condition, traffic information etc.• Self driving cars
CLOUD-CENTRIC ARCHITECTURECloud-centric Architecture rely on cloud resources to drive functionality and data management.
ADVANTAGES & DISADVANTAGESAdvantages DisadvantagesScalable as it relies on the computing power of the cloud
Requires a persistent Internet connection
Minimizes the cost, complexity and need for a user interface as the focus is moved from the things to the cloud
In case large amounts of data are to be transferred to the cloud then the solution becomes costly
All the benefits of cloud computing. In certain applications, security and privacy may be a huge concern so it might not be desirable to share sensitive data with the cloud
Centralized approach to network-level smartness
EXAMPLES1. Fleet management:
• Problem- Rising fuel costs, environmental concerns, driver safety and unpredictable maintenance problems.
• Solution- Vehicle data is collected using sensors and pushed to the cloud where this data is analysed. This helps in improving fuel efficiency, reduces maintenance costs and improve driver safety.
2. Parkinson’s Disease• Problem- Significant challenges in monitoring symptoms, the progression of the disease
and the effectiveness of drug therapies. • Solution- 300 observations per second are collected from each patient. This anonymous
patient data is aggregated and analysed. Machine learning and graph analytics deliver more accurate predictive models.
EXAMPLES3. Parking Analytics
• Problem- Efficiently managing parking spaces.• Solution- Parking lot sensors identify the presence of a vehicle, relaying the
information to the cloud. Provides possible intelligence to see what is happening on the streets and in parking lots to make data-based decisions.
ENTERPRISE-CENTRIC ARCHITECTURE
This architecture is focused on keeping connected machines, application logic, analytics and datastorage on-premises — that is, behind the enterprise firewall.
ADVANTAGES & DISADVANTAGESAdvantages DisadvantagesSecurity and privacy are better contained.Restricting Internet and public cloud access toonly that which is required helps an enterprise leverage the benefits of the IoT without actuallygoing on the public Internet.
It is not well-suited to situations where many machines are geographically dispersed. The coordination of the machines will require central oversight.
Real-time control and monitoring are enhanced if all resources are on-premises and in close proximity.So better control of resources.
The benefits of the cloud are lost.Cost efficiencies, scalability, resilience and other benefits of the cloud are not prominent in this architecture style.
All computing resources and storage must be on-premises.Which will increase cost and complexity.
EXAMPLES
A hospital can take advantage of connected things without actuallyhaving to go out to the public Internet. These enterprises are willing to give up the benefits ofthe public Internet for the added security and privacy of an on-premises deployment
CHALLENGES IN IOT
TOOLS FOR IOT
n.IoThingworxSensorflareThethings.ioZatarCrowsnest
DIFFERENCE M2MAND IOT PRACTICE
M2M is almost synonymous with isolated systems of sensors and islands of telemetry data. In contrast, the IoT is trying to marry disparate systems into an expansive system view to enable new applications — that’s not only the big idea, it’s the one key difference between M2M and IoT.
KPI AND CRITERIA
EVALUATION AND DECISIONCost
•Cost of hardware, software and data. Choose where the operational cost will be in your architecture. For example, putting a powerful processor or a large data store on an asset puts cost on the thing, but removes cost from the cloud or gateway. Cost applies to hardware, software, storage and communications costs.
Connectivity and Technical Requirements
• Reliability and quality of service. If the use case requires mission-critical response times and reliability, then choose the architectures that support this. For example, an on-premises architecture keeps things in-house and enables maximum control for
management and compliance.
• Real-time performance. The closer the applications or analytics are to the thing, the easier it is to get a real-time response. There are technologies and approaches that overcome this challenge (for example, high-performance messaging), but in general, a real-time response benefits from physical proximity to reduce latency.
EVALUATION AND DECISIONConnectivity and Technical Requirements
• Distance and physical constraints.Will the things be located far away, globally or on-site? If the things are located far away or globally, will they have problems connecting? Will the things be in a wireless-unfriendly environment? There is no avoiding physical
limitations when working with the IoT. Choose architectures that work around these physical limitations.
• Speed and quality of connection.Will the things need to be continuously connected to the Internet? Are they in a physical environment that allows for a high-quality connection? If connection is sparse, choosing a thing- or gateway-centric architecture is
preferable.
• Power requirements.Managing power consumption is an ongoing struggle in the IoT world. For example, putting a lot of computing resources on a thing (for example, a multi-core processor) makes the thing more self-sufficient, but also requires more local power. In contrast, making the thing as dumb as possible (for example, leaving only sensors and a microcontroller) may remove computing resources from the thing, but it also lessens power requirements.
EVALUATION AND DECISIONData and Security
• PrivacyThe usage patterns of things can be analysed and can lead to privacy issues. Think of a consumer products
company analysing how often you use your appliances to determine how clean your household is. Choosing to keep data behind a gateway, or having a healthcare enterprise opt for an "intranet of things" in each hospital, keeps privacy up.
• SecurityThe architecture styles chosen will give an enterprise an advantage (or disadvantage) in the challenge to secure the IoT. For example, choosing to keep data on the thing or the gateway minimizes the chance that the data will be intercepted. Adding cascading gateways or limiting IP addressability to the gateway (and not to the things
connected to the gateway) adds extra security. Many things will have too little of a technology footprint to have any decent security technology on-board. By relying on the power of gateways, smartphones or the cloud, enterprises can leverage more powerful security technologies than what the things could support on their own.
• Amount of dataSome things will send out 1 byte (or less) of data when a threshold is triggered. Other things will have huge
amounts of streaming data — to the point that it is impossible (in terms of technology or cost) to send the data centrally. The value of data is also an important consideration in terms of monetary value and security.
EVALUATION AND DECISIONUsers and Implementation
• Greenfield or brownfield implementation.Will you be building from scratch (greenfield), or starting with a legacy environment that will need retrofits and new technology? Most legacy connected-machine environments are already thing-centric or on-premises-centric.
• User needs.Will users want to operate the thing in a disconnected fashion most of the time, even though connectivity is
available? Will users want to run the post analytics on a plane or in the field where no connection is available?
• Integration needs.Determine what the things will need to be integrated with — it could be cloud storage, ERP systems or control systems. The right choice of architecture can simplify integration complexity. For example, some cloud-based IoT middleware technologies provide ease of integration: "Connect it to the cloud, and we'll take care of integrating it on the other end."
• Channel.As in Web, mobile, social networks and so on. Users needing to access data and functionality from a social network (for example, needing to have a machine tweet) or from a mobile app will draw you toward certain architecture styles.
IOT USE CASE – SMART HOSPITAL
CONCLUSIONUse the Five IoT Architectures as Models — To Be Combined and Changed as Required.
Use these steps as a way of thinking about how to implement these architectures: 1. Find the architectures that fit your use cases.
2. Choose or build an IoT platform that can support these chosen architectures — (ideally, allarchitectures, even the ones you won't adopt immediately).
3. Consider emerging technologies that may eliminate the advantages and disadvantages ofan architecture style.
Q & A
REFERENCEShttp://www.intel.com/content/www/us/en/internet-of-things/infographics/guide-to-iot.html
http://www.gartner.com.libproxy1.nus.edu.sg/document/3098434?ref=QuickSearch&sthkw=iot%20architecture&refval=156299155&qid=5bf77c003ec9eaa0298b4bba9ca7b544
http://www.ibm.com/analytics/us/en/internet-of-things/
http://www.slideshare.net/ValaAfshar/internet-of-thingsslideshare
http://www.slideshare.net/CiscoIBSG/internet-of-things-8470978
THANK YOU