AWS IoT Deployment Helped Customer Achieve 50% Effort Reduction

The customer engages in a wide range of petroleum activities, ranging from upstream exploration and downstream oil refining to shipping and automotive engineering. The customer aims at embarking on a strategic journey of Data Analytics, for supporting their business requirements and becoming a data-driven company. The customer wanted to enhance its transportation safety and optimize its shipping and transport facilities.

The Need to Set Up IoT on AWS Cloud

The customer wants to become data-driven to support their business objectives, which gave them an impetus to adopt Data Analytics. The proof of concept is to set up IoT to show the process of running video analytics on the edge using AWS services. The data platform will enable the customer to perform descriptive and diagnostic analytics on key metrics, building the ability to mitigate the bottlenecks in the business. The customer intends to use the power of their algorithms on the edge.

The customer was looking to reduce in-vehicle violation detection with the help of IoT devices. This involved

• Distractions

• Hands off wheel

• Off Seatbelt

• Using Mobile phone – Texting

• Drinking while driving

Key objectives of the customer were to

• Optimize their in-house algorithm in a preferred IoT device, such as Raspberry Pi or Jetson Nano.

• Develop trigger calls for defined actions.

• Deploy optimized in-house algorithm on AWS IoT Greengrass

Blazeclan’s Proposed Solution Using AWS IoT Greengrass

With a traditional network of edge devices, it is extremely difficult to track, monitor and manage connected device fleets. AWS IoT Greengrass provides a platform for the IoT devices to operate even with intermittent cloud connectivity with secure communication, integrate it with other AWS services for added functionalities, and deploy code to millions of devices remotely at once

The customer was testing out several devices that fit their performance criteria. The customer was facing challenges in deploying their machine learning (ML) models on the IoT device integrated with AWS cloud services. Our team had to study their existing system and devices and configure them to run their code on AWS cloud IoT. The approach involved:

• Optimization/modification of the code to make it Lambda-compatible.

• Deployment of the code and ML models through IoT Greengrass.

• Figuring out ways to use the device’s GPU and VPU to minimize dependency on the CPU.

• Optimization of deep learning models on OpenVINO to speed up the inferencing on Intel architecture.

• Adding functionalities to code for sending inference video files to S3. This will translate IoT Core as MQTT messages and store them in DynamoDB for further consumption.

• Automating the process of device registrations for bulk deployment on more than 200 trucks using IoT Device management in future.

Benefits Gained by the Customer

• Near Real-Time Response: Using AWS Greengrass, the company was able to build a system that can respond to local disasters quickly, even if the device is offline by using the cloud for management, analytics and storage.

• Communication Security: The data transmitted from the devices locally or to the cloud are encrypted and authenticated using X.509 certificates. This is to ensure that the data is never exchanged between devices and the cloud without a proven identity, thereby reducing the risk of man-in-the-middle attacks.

• Comprehensive Platform Support and scalability: The customer realized the flexibility to use a plethora of devices that can run AWS Greengrass according to their business needs. The serverless infrastructure has also increased agility and scalability.

• Facilitated Remote Management of Devices: The solution facilitated the customer in maintaining the health, managing, and deploying code on a fleet of devices. This further increased the control over managing the complete IoT network and defining how devices communicate with each other.

Tech Stack