Telematics : Fatter and Bigger
Telematics has slowly started to become the new industry talk point with the big data gaining momentum in-terms of high processing speeds, cheap storage capacity and easy to use cloud system providers. A lot of industries have started to tap this data and mine useful information from it. When we talk about scalability on servers we generally tend to talk about horizontal and vertical scaling. I see telematics scaling in both the ways, horizontally scaling, growing scenarios where telematics is being utilised in various industries, vertically scaling, within the same industry, more and more data points being collected with more and more sensors. In this blog post, I will be walking through some of the uses of telematics in Automobile industry. I will also walk through a general architecture which can be used across different industries in telematics.
Telematics is a term that combines the words telecommunications and informatics to broadly describe the integrated use of communications and information technology to transmit, store and receive information from telecommunications devices to remote objects over a network. As in automobile industry use cases, I will be describing Vehicle Telematics and Industrial Telematics to explain its usage in these streams. We can also call them Connected vehicles and Connected machines
Vehicle Telematics
Telematics have numerous use cases when it comes to analysing vehicle data. Before listing down some of the use cases, let me first go through a very simple design architecture of the data flow as shown in Fig 1. Sensors data is generally a time series data and can best be stored, processed and analysed as time series graphs.
- Data from various sensors travel from telematics unit, installed in the vehicle to the gateway servers.
- This data then passes to the data pipeline from the gateways, acting as a temporary data store for processing.
- Next, the data is moved to a time-series database for various analysis
- Time series dashboards can be attached to this database to chart out various graphs.
When you talk about telematics from technical standpoint, one thing to keep in mind while designing any architecture is scalability, as you will be bombarded with sensor data, therefore at every point, your technology should be scalable. For instance, at point 1, Api gateway servers can be designed to be auto scalable, at point 2, Data Pipelines are horizontal scalable using any open source like Kafka, at point 3, no-sql time series databases like Influxdb are scalable depending on load.
Listing down some of the scenarios where this data can be utilised but not limited to these
- Vehicle Insights
- Rider Insights
- Usage Based Insurance
- Fleet Management
- Predictive Maintenance
- Theft Identification
- Road Side Assistance and many more
Industrial Telematics ( IIOT )
If you compare Fig 1 and Fig 2, the backend architecture remains the same, only the data sources changes. Since all the data is time series, this basic architecture can be used across various telematics scenarios. It can again be extended to include other big data services like security, monitoring, ETL(s), data warehousing, big data processing and machine learning.
In a typical assembly line, there are various phases on assembly for a vehicle. Slowing/missing/defect in a phase can lead to production issues. Once we have sensors deployed at various phases, we can find out the issues as well even predict issues at a very early stages of the assembly.
Listing down some of the scenarios where data can be utilised but again not limited to these
- Predictive Maintenance
- Raw material usage prediction
- Engine noise Analytics and many more
This is the first part of the series on telemetry. I will be going through each one in more detail in the coming parts. Let me know if you find the information useful. I would love to hear your feedback.
Edit 1
Please go through this link vehicle telematics for the second part of this series
Connect with me on Linkedin
