Professional Documents
Culture Documents
UCT IoT Deployment and Challenges FDP
UCT IoT Deployment and Challenges FDP
About Myself
• IIT Kanpur Alumnus
• 19+ year Industry experience
• Presently working as Director at UniConverge Technologies Pvt. Ltd.
(UCT)
• Chief Mentor of The IoT Academy “Skill development and Trainings”
• Worked with many MNCs
• Ericsson AB, Sweden (for 7 years)
• STMicroelectronics Pvt. Ltd, Noida
• UbiNetics India Pvt Ltd, Bangalore
• SASKEN Communication Tech., Bangalore
What to discuss today?
• Glimpse of a real usecase
• IoT Layered Architecture
• Design Challenges in each layer
• Sensor selection
• Gateway selection
• Connectivity selection
• IoT Cloud platforms
• Deployment Challenges
• Best Practices
• Today's Takeaway
• Q&A 3
IoT is not a technology,
but
It is a complex ecosystem
4
Glimpse of a real usecase
5
Indian Railways Scenario (Northern railways case)
Indoor Signals (Relay
Room)
8
9
IoT Layered Architecture
10
Future Development would be around “DATA”
11
Technologies around “DATA”
• From where Data come from : IoT/ Embedded
• How data is communicated from source to destination : Networking
& Communication Technologies
• How and where data is managed : Cloud Computing/
Big Data
• What to do with Data : Machine Leaning
• How to secure data : Cyber Security
12
IoT Architechture
Challenges
14
Challenges
• Design challenges
• Deployment challenges
• Security
• Scalability
15
Design Challenges in each layer
16
Design Challenges
• Sensor selection
• Gateway selection
• Connectivity selection
• IoT Cloud platforms
• Basic features
• Security
17
18
Functional requirements of IoT Platform
Data collection from the devices over different protocols and network topologies,
Device Registry and management
User and Access Management
On-device data processing (Edge processing)
Data Representation and Visualization
Reporting, Rule Engine, Alerts and Notification
Data processing and BI
Components for frontend analytics
Cloud-based deployment
Security : TLS/ Encryption
Remote device configuration and control
Over-the-air firmware updates
Scalable (cloud native) / Load Balancing
Customizable
19
Integration with 3rd party software
Performance requirements of IoT Platform
• Scalability
• Stability with increase of Data volume
• Ease of use
• Good User Experience
• Easy deployment
• MQTT Performance (How many devices per client thread)
• Multi threaded
• Multi Client
• Multi topics
20
IoT Cloud Platform
21
AWS Cloud Platform
22
GE Predix Cloud Platform
23
Databases
25
Properties of IoT Data
26
Database Requirement for IoT
27
Evolution of Database Management System
28
Simple Database integration
Device Registry, Mgmt and
Server Config.
Predictive Maintenance
Business Intelligence
&
Third Party ERP APPs
29
DB Selection
• What is good choice of DB at server?
• How can we make DB load and multiple access agnostic?
• Data queue management tool e.g. Kafka etc
• Scalability and Stability
• How many devices/channels/TAGs can be supported?
• In case of Edge computing,
• Huge volume of IOT Data acquisition, filtering, storage in DB
• How DB is managed for high volume?
• Highly loaded access of DB
30
Opportunities
Challenges
Challenges
• IoT suffers from Market Failures and Externality Effects e.g., Accrual of
societal benefits depend on the broad adoption of IoT-enabled solutions
(thermostats/health). E.g., Competitiveness Externalities
• IoT suffers from “Chicken-and-Egg” Challenges e.g., The success of many
IoT apps depends on the success of complementary technologies/systems
(e.g., energy/transit).
• Interoperability and Standards-Setting Issues e.g. Governments have an
important coordinating role to play in developing large-scale deployments
of sensor networks and smart infrastructure that spans multiple
jurisdictions.
• Ensuring adequate radio spectrum and coherent regulatory approach.
Challenges
• Global cooperation
• Proprietary and incompatible protocols
• Lack of APIs
• Example: Common external power supply
• Technological challenges
• Power usage
• Scalability
• Security
• Communication mechanisms
• Ethics, control society, surveillance, consent and data driven life
IoT Data- Challenges
• Multi-modal and heterogeneous
• Noisy and incomplete
• Time and location dependent
• Dynamic and varies in quality
• Crowed sourced data can be unreliable
• Requires (near-) real-time analysis
• Privacy and security are important issues
• Data can be biased- we need to know our data!
35
IoT Data Challenges
• Interoperability: various data in different formats, from different sources
(and different qualities)
• Discovery: finding appropriate device and data sources
• Access: Availability and (open) access to resources and data
• Search: querying for data
• Integration: dealing with heterogeneous device, networks and data
• Interpretation: translating data to knowledge usable by people and
applications
• Scalability: dealing with large number of devices and myriad of data and
computational complexity of interpreting the data. 36
Typical Industrial Challenges
Challenges: Connectivity's Diversity
Challenges: Protocol's Diversity
Deployment Scenarios: Industrial Automation
Deployment Challenges
41
Response
• Sampling of Analog Signals < 100ms
• LPWAN gives better Range (upto 3km in Rural) in case of line of site. In case of
No LOS, range may vary from 500 meters to 2000 meters.
• Measurement quantity will carry server time stamp that is sync with device time.
• If analog signals are concentrated then they can be sent over Modbus, Gateway
can handle upto 100 Analog signals but if Modbus is not possible then data
needs to be sent over LPWAN to Gateway, that can handle 10 Analog signals
(Need to write correct figures, check original question)
• Data is sent to server only when it exceeds more than preset threshold
variations.
42
Challenges
IoT Solution
Analog Analog to RS485
Signal converters DAMS HW
Sources (8AI to RS485)
Modbus over Cloud Server
Air Inteface
RS485
Multiport
* Ethernet to
* DAMS HW 2G/4G Modem
*
Ethernet
DAMS HW
CASE 1: CASE 2:
No. in order of 10s No. in order of 100
Yes
Yes
Is Internet
Connectivity No Use 2G/4G for Cloud
Available over Server Connectivity
Yes Ethernet?
Use Ethernet for Cloud
Server Connectivity
If End node are No
clustered and close by
47
Information Security
Office of Budget and Finance
Education – Partnership – Solutions
Why be concerned about IoT?
• It’s just another computer, right?
• All of the same issues we have with access
control, vulnerability management,
patching, monitoring, etc.
49
What Can Breach IoT Networks?
• What can’t?
• Billions of connected devices
• Secure and insecure locations
• Security may or may not be built in
• Not owned or controlled by IT … but data flows through
the network
56
Information Security
Office of Budget and Finance
Education – Partnership – Solutions
Recommendations
Accommodate IoT with existing
practices:
• Awareness Training
• Risk Management
• Vulnerability Management
• Forensics
Information Security
Office of Budget and Finance
Education – Partnership – Solutions
Recommendations
• Plan for IoT growth: