# 5. Tech architecture

### Minimizing IoT Data and leverage blockchain based storage&#x20;

Blockchain based storage solution can help prevent below issues and extract max value:

1. Centralised data storage > data security and availability of service issues&#x20;
2. Ownership of data > misappropriation of data without owner permission
3. Privacy> infringement of IP protection and disclosure of sensitive information
4. Cost vs Incentive of sharing data > upkeep of data becomes cost instead of benefit to data owner
5. Customisation of data structure > various data structure and stream without standardization <br>

#### Minimize data before upload by extracting and sending only the context = metadata

<figure><img src="https://3856084885-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FzvVPkYkZrDhkSJg7GYyv%2Fuploads%2Fnv5ZeNU0PuUk3RUlwzUD%2Fminimizing%20iot%20data%20and.png?alt=media&#x26;token=ebd0d8f1-222f-4802-a0f0-e44a5f27cee2" alt=""><figcaption></figcaption></figure>

### AI in IoT

AI is based on Deep Learning algorithms. Deep Learning involves automatic feature detection from data. Data types in IoT and techniques used in each are:

<figure><img src="https://3856084885-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FzvVPkYkZrDhkSJg7GYyv%2Fuploads%2FhtrYlg7LQZ2YR0RkaRLN%2Fai%20in%20iot%20.png?alt=media&#x26;token=38ccc1c0-0721-471f-b99a-b6485fd39b90" alt=""><figcaption></figcaption></figure>

The difference between other fields and IoT is the volume of data and need for sophisticated real time implementations of the same models. T

Usage of the models or derived solutions are different for each IoT verticals.

<figure><img src="https://3856084885-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FzvVPkYkZrDhkSJg7GYyv%2Fuploads%2FOHdPbv9tfwCqymbWnjlg%2Fai%20in%20iot1%20.png?alt=media&#x26;token=27e7822f-36f6-4ac7-81c5-8a4e12cc0997" alt=""><figcaption></figcaption></figure>

WHAT IS COMMON both vertically and horizontally is the fact that <mark style="color:red;">large amount of data is required</mark> to derive these models.

\ <br>
