Fog_os

Fog Computing the Path Ahead for IOT

ManagementTeamMouli1

Kavitha Gopalan

The growth of IOT is explosive and formidable. Connected devices are becoming part of everyday life, from connected devices at home like the refrigerator, coffee maker, thermostat, bulbs, to connected vehicles, street lights, parking system, garbage can on roads to connected tools & machines in the factory to Connected equipment in a hospital.

In typical IOT architecture, the sensor data from connected “things” are streamed to the cloud, this data is stored in the cloud and then analytics and machine learning applied. According to industry research firms like Gartner and IDC, the IOT adoption is expected to grow significantly from around 5 billion devices in 2016 to 50 billion devices by 2020. Imagine all these devices streaming data to the cloud. How much volume of data would that be? According to Cisco, the data generated from connected device is expected to scale to 507.5 zettabytes (1 zettabyte = 1 trillion gigabytes) data by 2019.

If the reliance on cloud computing, as the foundation for IOT is continued, then this massive volume of data could throw new challenges like latency, bandwidth consumption, security and so on… The whole purpose of IOT could be defeated if there are latencies in delivering data. A connected smoke alarm sending a few seconds delayed message of smoke detection is of no use, the computing and decisions should be real time. This becomes more critical to the Industrial scenario where several sensors attached to devices need to send data to cloud for processing and storage. Real-time feedback and action are important if an impending failure is detected. Any latency or bandwidth issue could lead to huge impact

The solution is to bring the data and analytics closer to the edge, this can significantly increase the response time of the IOT systems. The edge part of IOT is where the sensors and devices are. Fog computing is bringing the computing power closer to the edge.
The key goals of fog computing are

  1. push computing to the edge so that not every data is sent to the cloud
  2. Intelligence build in fog to take action based on data analysis
  3. Send selective data to cloud

Sensors are typically legacy devices and they collect data, there are no storage or intelligence build in. And every sensor will have their own communication protocol. The way to enable fog computing is to have fog nodes or gateways that are capable of receiving the data from several sensors. They should also be capable of storing this data locally and run analytics for any real time action. The fog node software should be capable of sending the selective data to the cloud.

PAASMER IOT platform architecture implements this kind of hybrid approach where the computing and intelligence are pushed to the edge at the same time sending selective data to cloud for recovery storage and enhanced analytics. PAASMER software stack – operating system and middleware sits on the fog node or gateway allowing connectivity to sensor following any protocol, Ingesting data, and process. There is light weight analytics engine in the software that provides the intelligence to the fog node to dictate action in real time to the sensor. The middleware software also allows data filtering and backend integration to the cloud for subsequent data storage and deep insight analytics.

To know more visit us at www.paasmer.co

Leave a Reply

Your email address will not be published. Required fields are marked *