Above listed items appears to be using AI and providing successful services in various segment. All of them embedded, but cloud connected applications.  May be car and satellites will have less dependency on cloud side support  and carry out most of required inference on board.

Most of the above listed items, using GPU based massive parallel computing on board for real time inference.  Trained Deep learning networks deployed on each system for inference and GPU based CUDA cores are used for parallel computing requirements of Deep learning networks.

FPGA and GPU are competing  devices in massive parallel computing requirement.  It appears that Jetson series ( Nano, TX2, Xavier, Orin etc) had opened up huge advantage for IoT edge design and development teams across the Globe.  Where Jetson series devices are from Nvidia with software support via CUDA SDK and associated libraries.   

FPGA also playing critical role in embedded devices where input reading is necessary and input reading via JESD204B or  Serial LVDS.  In the case of Jetson series, input reading  ( high throughput ) is possible via PCI gen 4 .   

May be soon, there will be a standard for high throughput reading from sensors for real time inference.

Case Study : Deploy AI application in IoT Edge computer

Step 1:  Run MQTT client in IoT Edge

In this case Jetson Nano is used. 

Local support will be given  to run python version of MQTT client to work with IBM Watson IoT platform. 

Step 2:  Run TCP/IP socket Server to connect with IoT Nodes 

Embedded board  is working as  IoT node.  ARM CPU is used in IoT Node.  Local support will be given in the form of Sample code  for IoT Node.

Step 3:  Run DC motor Control app in IoT Node

Embedded board with ARM cpu is used as  a IoT node. 

Local support will be given in the form of Sample code  for IoT Node.

Step 4:  Create IoT device in IBM watson IoT platform 

Local support will be given  to create IoT device in ibmcloud.  But user need to have account in , in case not then new account will be open during event for each user in

Step 5:  Create App in  Node-Red

 Local support will be given  to create app in Node-Red in ibmcloud 

Step 6:  AI Application in IoT Edge

Run Image Classification app in IoT edge  and run a motor if given image is Banana.  And also update MQTT broker in IBM watson iot platform such that MQTT broker can publish this news on Banana to app in Node-Red

Amount of AI deployment in IoT Edge segment is growing quickly.  IoT Edge, Near Edge and  Embedded Systems are sharing similar constraints on available computing for application which is native to Edge.  

Questions from IEEE member after Presentation "AI in IoT"

Q 1. Would be interested in Embedded intelligence and IOT concepts as well as how it would help me in designing embedded product better.

Ans :  Domain knowledge is key ingredient to deploy intelligence in embedded  application. With minimal effort on Machine learning skill development will enable you to handle and design service  ( M/L on it)  that runs on embedded platform.   

Q 2. Want to know more about iot and electronics things

Ans :  IEEE offering 3 courses on IoT. These 3 courses will make you to feel confident on IoT and also associated Electronics 

Q 3. Using IoT in current working environment.

Ans :  Super. Well set

Q 4. Use of iot in projects related to telecommunication

Ans : Telecommunication is playing major role in connectivity. For example upcoming Narrow Band IoT version LTE, appears to be very good for connected edge to communicate with Cloud applications.    ( )

Q.5 Security issues in IoT End Devices

Ans :  may be Elliptic curve based  version will be very useful  ( for example  Y^2 = X^3 + 7 ) 

Q 6. Scope for use of

Ans :  Mini and Micro payment part appears to be coming up.  

2) Agriculture

Ans :  Emerging 

3) Water testing

Ans :  home water management appears to be interesting 

4) Security of IoT

Ans :   Refer Q .5

Q 7. Practical challenges and Risks

Ans :   Post deployment ( to upgrade embedded version of IoT edge and Node) appears to be costly issue.  The same might define service quality.

Q 8 Network simulation for a smart grid

Ans :   Case study is given in presentation 

Q 9. Knowledge on embedded coding and on some basic electronic devices.

Ans :   C level coding will be helpful.  Board level support package development also helpful. On electronics side, Data sheet handling will be essential while creating BSP.

Q. 10 IoT penetration in the social sector (specific cases)

Ans :  wearable version will be playing major role in social network.  Case study is given in the form form Russel character in presentation.


 Q. 11 IoT backhaul

Ans :   mostly MQTT broker platform emerging at backend to support Query in the form of PUB-SUB model. Narrow Band IoT ( LTE version) appears to be emerging as main contender.

Q 12 IOT application in Electrical fields and areas

Ans :  case study is given in presentation.

Q 13 Information regarding 5G

Ans :    Amazing question. IEEE standards group need to take this question. However, there is Narrow band version getting in 5G for IoT services


Q 14. I want to learn sensor integration, Wireless communication and how to use our algorithms in CLOUD for IOT smart building application.

Ans :   Sensor integration with Host CPU board can be done by using UART, SPI, I2C etc.  IEEE IoT course on Sensing provides information to learn the same.  

           Wireless Communication

Q 15. I want to know more on edge and fog computing

Ans :   Edge is close to device or sensor network and also actuar network.  Fog computing is more like a on premise cloud computing infra,

Q 16. I need to know more about energy efficiency  and Localization in IoT

Ans :  Sensing , computing and communication are three parts apart from control segment at edge.  PUB-SUb model appears to be resulting in less bandwidth requirement and also intelligence in Edge will result in ultra low bandwidth requirement to reach out cloud side of computing infra.

Q. 17 Http communication between IoT devices.

Ans :  Restful using HTTP to provide service  between  2  terrminals that are connected via TCP/IP network.   GET, POST are few popular methods of Restful.

Q. 18 How to set up and work with Wireless sensor networks

Ans :  IEEE 802.15.4 provide std to setup Sensor network ( Zig-bee is one such a version on top of Layer-2 ). Application need to run on Layer-2 to work with sensor networks such that multi hop can be achived. 

Q .19 How the fourth industrial revolution will change the world and what kind of billion dollar company will be emerged in this revolution?

Ans :  Industry 4.0 is the hello world of new generation automation in upcoming manufacturing companies. For example, FANUC is providing robts that can handle Electronic subsystems assembly with ease. This results in lower cost of manufactruing in EMS companies.  It appears that IoT going to bring  small and micro compaines as Assembled PC kind of market ( mid nineties ) .  But still there will be few companies emerge as 100 plus Billion USD ( from Semiconductor segment)

Q.20 How Edge solutions help in IoT to offer better service ?

Ans :  real time and useful for applicaitons like ADAS.  

Q. 21 Hardware platforms for IOT implementation

Ans :  Sensor ( or sensor network)

          Local Host CPU with memory and network interface ( BLE, WIFI etc)

          MQTT broker service hardware

          Cloud ( example  IBM Bluemix, IBM Watson IoT service etc)



Ans :  IoT deployment is happening in many sectors  ( healthcare, factory floor, transport, oil and gas sector and many more)

Q. 23 Edge computing / AI at Edge need a handson Guide to how to cultivate this skill

Ans : IEEE IoT course is available ( via BLR)

Q. 24 distributed neural networks

Ans :  ANN is one of the best Machine learning option. However distributed version of ANN might require high speed connectivity.

Q. 25 Data management in IoT

Ans :  Blockchain emerging and we need to wait and see.  Every IoT device ( edge or node) will have its version of data in cloud. ( history as well as current data)

Q. 26 Data Analytics and IoT integration

Ans :  data analysis done at  three level. 

           First analysis done in sensor while collecting data

           Second analysis done in IoT edge by using machine learning ( AI)

           Third analysis done in cloud ( for example, IBM WATSON based services such as conversation, 

           NLP< discorvery etc) .

           Every IoT Node and IoT Edge will be have its version in Cloud 

           ( for example in IBM WATSON IoT service.)

Q. 27 Code to implement the function and for integration

Ans :  IEEE  IoT course via BLP will provide good amount of handholding on the above.

Q. 28 Cloud computing

Ans :  IBM Bluemix is one such a cloud platform 

Q. 29 Building Automation using IoT

Ans :  covered in presentation.

Q. 30 Being an amateur, I want to learn enough to get an understanding of the key things involved and to start working on applications.

Ans :  IEEE  IoT course via BLP will provide good amount of handholding on the above.

Q. 31 As a mechanical Engineer i want to know how a controller need to be selected, what is protocol and how it works, how AI can be effectively utilised

Ans : IEEE  IoT course via BLP will provide good amount of handholding on the above.

Q.32 Any running technology like cloud computing and big data

Ans :  Both are managed in cloud side. For example Discoovery service of IBM WATSON provides real advanatge to mange  big data from various segment. For example it can watch movie and provide a short version of video as a trailer. Many more can be managed by IBM Watson IoT service.

Q. 33 analytics on IOT data

Ans :  Intellignce in Edge provide some amount of analytics on IoT data. But mostly done in cloud side. For example IBM WATSON IoT service provies many paid service th handle analysis on IoT data.

Q. 34 Acknowledgement of Intelligence with concern to Strong AI.

Ans :  AI is good to use in IoT service. Concen need to be addressed on validity of data and owner of data.

Q. 35 About iot so that i can do my final year projects on it

Ans :  IEEE  IoT course via BLP will provide good amount of handholding on the above.

Q. 36 About IOT and any difference between AI and EI?

Ans :  AI is getting used in IoT service. For example, IBM WATSON Iot Service provide service in AI . In the case of EI, there is need to create thinking Robot by using cognitive intelligence. May be soon and it might come up well along the research in ADAS. 

Q.37 About internet of things and their use in day to day life

Ans :  Good question. Character Russell is designed and presented in webex session.

Question from  Shri Rajesh K Jeyapaul


Question :  good to know it a quantized model ?


super question.

Maybe I can take you along to www.epigon,in and its innovation in similar problems.

For example, a music decoder ( mp3)  produces music if a computational workflow is applied on a given encoded  mp3 bit stream.

ISO /IEC 11172 - 3   does provide a computational workflow for mp3 decoding.  But there is no guideline for executing that in a given DSP or CPU.  Thus experienced engineers who understand  DSP / GPU well are required to carry out translation of ISO IEC  given computational workflow into feasible work for a given DSP  or CPU. This 8s trick that results in writing an mp3 decoder for a given DSP.   Decoded music might be similar and need not be same. Thus quality of audio depends on 16 bit dsp vs 24 bit dsp etc..


Question from Shri Jagadeesha chinagudi 

Researcher / Scientist / Engineer at Indian Space Research Organisation,NRSC-RRSC-South

Question : Highly stuffed informative message. The practical applications in the field departments and industries like MSME are yet to begin digitization process ,digitization process or digital transfer process. Any information on these for Indian Govt Departments and MSMEs are available? 


Sir gm.. thank you so much for your kind words.


Sir ,  Digital transformation is about entering MSME and also in Govt Departments. Mostly , the plan is to deploy in a local server machine or in an Amazon cloud machine. For example, camera is collecting pic and inferencing done on the server.  Camera is just an image collection device and there is no AI inference in camera.   Imagine, companies deploy a camera with AI.!!!!

sir ,we can give this IP  to  companies  and let  them make their staff  add value instead of starting from the basics.  For example , their work force can use this IP to make AI camera or AI cooker or AI street and AI in small office security..etc..

Question : I think, I  have to clarify some terminologies,  I used earlier by me Digitization means e- records ( doc or docx or  odt or Excel sheets etc) of the MSME companies or even a gram panchayat or Watershed department ( it's sub divisions) or horticulture department (it's sub divisions. ) Irrigation department or revenue department , land records department etc 

Digitization. :I mean arranging e- records or event  adding new records to understand the processes of the management of office works mandated to them, 

Digital transfer means - The CSE and ICT technologies needed to automate the processes of an organisation to deliver its  mandated functions .For ex: Farmer producer organisation which has its functions as e- telling the farmer the price of agri products, availability ( geographical locations )  of seeds fertilizers pesticides, tractors, harvesters , cold storage transport infrastructure nearest to him etc. Other exams is say water supply network : Automated water supply system at village s / gram panchayats should transfer all its network indicators on to a dash boards at Panchyat level to know at what amount and velocity and pressure both drinking water is supplied to all households daily or certain time intervals depending on availability of water , ground water and tank water use for agriculture as a process of functions under various institutes connected with gram panchyat departments  etc


Sir , AI work flow is part of Digitization and IoT also part of Digitization. DLtrain designed to provide deep learned Neural network   enabled real time ( of non real time inference ) inference capability.  For example Panchayat water network can deploy DLtrain in their  village water network to handle Quality of Water supply and  predict Peak requirements of water in a given day such that  people get to have high quality water supply service for them.  But use is , how to deploy AI in Water Pump, AI in Level sensor etc.  In above given tutorial,  Methods and apparatus are given in detail to handle "deployment of AI in Water Pump, or AI in Level Sensor etc".