distributed computing in iot

Distributed Computing in IoT System-on-a-Chip for Smart Cameras as an Example

By: Kishan Patel 26th October 2016

Contents • Introduction

• Challenges

• Solution

• Conclusion

• References

Internet of Things

• Sensors

• Communication

• Computation

• Service

• IoT is a collection of connected devices and services that work together to do useful stuff.

Components of IoT

Introduction

• In IoT there is billions of smart objects, such as

Sensors, Actuators, Smart phones and Smart

Vehicles, etc. are connected with each other to

sensing physical signals and giving better service

in real time human need.

• In this paper presentation, my main focus is on

sensors, communication & computation.

Challenges

• Ultra-Big data

• Infeasible and inefficient to

handle with cloud services

because of computing and

communication recourses.

Challenges

• Using centralized approached

where all the data analysis works

are executed on cloud servers.

• Cloud server have to do job of

Computing and Communicating

that processed data itself.

Solution • To handle Ultra-big data, Distributed computing is a

efficient way rather then centralize.

How Distributed Computing solve the problem?

Distributed smart camera

as a node of an IoT

Example: Distributed Computing In Video Sensor Network

Video Surveillance

• In this case target is to store video sequence when critical event occurs.

• Approaches: 1) Centralized approach: All data are

streamed back to the server. 2) Distributed approach: Intra processing

stage and Inter Processing stage

Example: Distributed Computing In Video Sensor Network

Video Surveillance

• As showing in figure, with Distributed Approach, 91.3% of the transmission bandwidth can be saved.

Example: Distributed Computing In Video Sensor Network

Vehicle Localization

• The Second case is a Vehicle neighboring map generation system with video cameras in an intelligent transportation system.

• Aggregator: • RSU (Road site unit): • OBU (On-Board unit):

Example: Distributed Computing In Video Sensor Network

Vehicle Localization

Here six different cases taken to test case:

• Low-end sensor is the video sensor with simple video capturing, coding and transmission ability.

• Middle-level sensor is the video sensor with moving object detection ability. It can transmit video data to aggregator only when it detects moving object.

• High-end sensor is the video sensor with a vehicle detection subsystem. It can transmit vehicle information instead of sending video data to aggregator.

Example: Distributed Computing In Video Sensor Network

Vehicle Localization

Result is shown in figure,

Sensors Corresponding Aggregators

All aggregators Cloud severs

Conclusion

• Distributed Computing is an essential technique for

internet of things (IoT) to off-load the computation

from the cloud servers as well as reduce the

transmission bandwidth requirements.

• Experimental results show the proposed design can

achieve high area and power efficiency.

References • K.-W. Chen, H.-M. Tsai, C.-H. Hsieh, S.-D. Lin, C.-C. Wang, S.-W.Yang, S.-Y. Chien, C.-H. Lee, Y.-C. Su,

C.-T. Chou, Y.-J. Lee, H.-K. Pao,R.-S. Guo, C.-J. Chen, M.-H. Yang, B.-Y. Chen, and Y.-P. Hung,

“Connected vehicle safety science, system, and framework,” in Proc. 2014 IEEE World Forum on

Internet of Things (WF-IoT), Mar. 2014, pp. 235–240.

• Shao-Yi Chien, Wei-Kai Chan, Yu-Hsiang Tseng, Chia-Han Lee, V. Srinivasa Somayazulu, Yen-Kuang

Chen, “Distributed Computing in IoT: System-on-a-Chip for Smart Cameras as an Example ”