Edge Device IoT Data Packet Splitting for Latency Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The increasing use of IoT devices leads to network latency and computation latency issues when processing data from multiple devices, especially with high-payload data like live video feeds, due to the need for real-time data transfer to edge devices and cloud servers, which can result in significant delays and bandwidth costs, and existing solutions either require costly hardware or minimal processing before data is sent to the cloud.

Innovation Solution

Implementing an edge device architecture with a pipeline that processes and derives insights from IoT data by splitting data into packets, using multiple instances of a second physical processor to apply learning models concurrently and annotate data packets, allowing for real-time processing and reducing the load on cloud servers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If data from multiple IoT devices is sent to edge devices and cloud servers for processing, then data processing capability is improved, but network latency and computation latency increase

Engineering Contradiction:
Improvedata processing capabilityVSAvoidnetwork latency and computation latency
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent segments the data processing function by introducing multiple worker threads that operate in parallel on different data packets simultaneously. This threading architecture divides the monolithic processing task into concurrent sub-tasks, enabling the system to handle multiple IoT device data streams without sequential bottlenecks, thereby improving processing capability while reducing latency

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements preliminary action through pre-allocation of worker threads and pre-establishment of processing pipelines before data arrives. The system proactively creates and configures the processing infrastructure in advance, so when data packets arrive from IoT devices, they can be immediately processed without initialization delays, reducing computation latency

Inventive Principle:
Principle #10Preliminary action

2Loss of energy

If data is processed in real-time at the edge device, then network bandwidth usage is reduced, but device complexity increases

Engineering Contradiction:
Improvenetwork bandwidth usageVSAvoidedge device complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the edge device into specialized functional components: data receivers for incoming IoT streams, packetizers for data segmentation, worker thread pools for parallel processing, and transmitters for outgoing data. This modular segmentation allows real-time processing capabilities to be distributed across manageable components, reducing overall system complexity while enabling bandwidth-efficient local processing

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary layer in the form of worker threads that act as mediators between data reception and data transmission. These intermediary processing elements handle the complex real-time analysis tasks, isolating the complexity from both the data sources (IoT devices) and the data destinations (cloud servers), thereby enabling bandwidth reduction without directly increasing the complexity of endpoint systems

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If multiple worker threads are used to process data packets concurrently, then processing speed is improved, but synchronization overhead increases

Engineering Contradiction:
Improveprocessing speedVSAvoidsynchronization overhead
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent employs copying by creating independent copies of the worker thread processing logic for each concurrent thread. Each worker thread maintains its own copy of necessary processing state and data structures, eliminating the need for complex shared memory synchronization. This copying strategy enables high-speed parallel processing while minimizing synchronization overhead, as threads operate on independent data copies rather than sharing mutable state

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3396924B1Real-time processing of IoT data
Publication Date: 2020.10.07 HEWLETT PACKARD ENTERPRISE DEV LP
  • EP3396924B1 patent drawingFigure 1
  • EP3396924B1 patent drawingFigure 1A~1B
  • EP3396924B1 patent drawingFigure 2A~2B

AI summary

Examples provided herein describe a method for real-time processing of loT data. For example, a first physical processor of an edge computing device may receive a set of data from a first loT device communicably coupled to the edge device. The first physical processor may split the set of data into a set of individual data packets. A second physical processor of the edge device process the set of individual data packets by: concurrently applying, by a plurality of instances of the second physical processor of the edge computing device, a learning model to each of a corresponding plurality of data packets from the set of individual data packets; and annotating, by a subset of the plurality of instances of the second physical processor, a corresponding subset of the plurality of data packets with a corresponding output from the concurrent application of the learning model.