edge computing a part of iot - altera.com · edge computing—a part of iot hidenori yakushiji,...
TRANSCRIPT
Edge Computing—A Part of IoT
Hidenori Yakushiji, Assistant GM New Biz Development Grp., FUJISOFT Inc.
1
2
Agenda
• Role of Edge Computing
• SoC FPGA Solution for Edge Computing
• Edge Computing GW and IoT Solution
• Summary and Next Step
3
Role of Edge Computing
Web Services
Search
Social Networking
Service
Voice/Image recognition
Medical/ Healthcare
M2M
Augmented Reality
Machine control
Da
ta V
olu
me
Real-Time
Cloud Computing
Edge Computing
NTT News Release: http://www.ntt.co.jp/news2014/1401e/140123a.html
Edge Computing Accelerates Response Critical Market
Gaming
Smart City
Smart House/ Building
Storage
Intelligent Transport
System
4
Role of Edge Computing: Value
… Technically, no question. Monetization is the key to balance profits vs. data value
• Upload data downsizing
• Response time
• Absorbing several I/F
• Security and manageability
• Flexibility and upgradability
• Cost
5
Agenda
• Role of Edge Computing
• SoC FPGA Solution for Edge Computing
• Edge Computing GW and IoT Solution
• Summary and Next Step
6
SoC FPGA Solution for Edge Computing
Optimization using SoC FPGA can offer the best of class cost/ performance on edge computing
Current Edge Computing Solution Next Edge Computing Solution: SoC FPGA
Power
Performance
Cost
7
SoC FPGA Solution for Edge Computing: Case Study
SoC FPGA Offloading 1000 : 15
Data Cleansing Filtering
Transform
: :
Valuable Data (<0.08TB/Day)
Sensing Data (5.4TB/Day)
8
• Extract specific frequency data
• Perform data cleansing
• Minimize data size
1. Re-sampling for fast Fourier transform (FFT)
2. Singular data filtering 3. Window function for FFT 4. FFT
:
SoC FPGA Solution for Edge Computing: Case Study Details:
Effective Approach : FFT Processing by FPGA
• Vibration data: 10K points/sec
• Getting multiple frequency factors by frequency analysis
• Peaks of specific frequency factors
9
SoC FPGA Edge Computing GW Enables Real-Time, Customizable Configuration Usage
SoC FPGA Solution for Edge Computing: Advantages of SoC FPGA Edge Computing GW
Update/Modify System Cost
Initial Cost Physical I/F update
Standard SW Hardware Processing
Capacity
〇: Real-time,
customizable, changeable specification ×: Initial cost
〇: Real-time,
low cost, ×: Fixed specification
〇: Standard software,
Initial cost ×: Real-time
Server
Edge Computing GW
ASIC/ASSP
〇 〇〇 〇
X
〇〇 XXX
X
〇〇 〇 X
〇 〇 〇〇 〇
〇
△ 〇〇〇
X 〇
〇
△ 〇
〇〇〇〇
〇〇〇 X
XX XXX
△
CPU
〇: GOOD △: FAIR X : BAD
SoC FPGA
10
Agenda
• Role of Edge Computing
• SoC FPGA Solution for Edge Computing
• Edge Computing GW and IoT Solution
• Summary and Next Step
11
Worker’s mental and vital healthcare is a critical issue
Edge Computing GW and IoT Solution: Demo: Target usage model
12
Demo: FUJISOFT* Can Offer Total IoT Integration
Edge Computing GW and IoT Solution: Demo platform overview
Public Cloud
Services
13
Join in the FUJISOFT* project development and trial! FUJISOFT offers custom features at minimum initial cost
Edge Computing GW and IoT Solution: Demo: Target Spec. of SoC FPGA Edge Computing GW
Edge Side I/F • USB 2.0 host • RS-232/422/485 • CAN • DIN/DOUT • Wi-SUN* • Bluetooth* LE • EnOcean*
• ZigBee*
Cloud Side I/F • Gigabit Ethernet • Wi-Fi* (b/g/n) • WAN (LTE/WCDMA/HSPA)
Operating System Platform • Embedded Linux*
• Other operating system
Assist Hardware • GPS • UPS • Extension slot • Encryption chip
Additional Feature • Robust security • Communication with
low latency • Rapid data analysis • Autonomy works at a
communication break
SoC FPGA
14
FPGA as the hardware offload engine FUJISOFT* offers custom features at minimum initial cost
Edge Computing GW and IoT Solution: Architecture
Embedded Linux*
Device Interface Protocol
Poster (Cloud Connector)
Data Analysis
Network Service
Data Conversion /Cleansing
USB
RS-232C
Bluetooth* …
MQTT
HTTP
Web Socket
Web Server
…
Security (ex. Cloud VPN)
Storage (ex. Embedded SQL)
Se
nso
rs
Clo
ud
FPGA Hardware Offload
(ex. IP-Core, OpenCL*, Original Logic) SoC FPGA
15
SoC FPGA Solution for Edge Computing: Vision of SoC FPGA: Genetic IoT Solution
• Balanced loading • Self feedback • OpenCL*
• Hardware re-program
• Frequent update : :
SoC FPGA: Re-configuration optimization through deep learning
A.I Self Optimizing Genetic System
SUM
Deep Learning • Accelerate processing through FPGA offload • Load separation and parallel processing through edge computing GW and Cloud
16
FUJISOFT*: IoT Integrator from Design to Cloud
Edge Computing GW and IoT Solution: FUJISOFT Solution and Plan
AI
Total Support of Edge Computing GW Solution 1H 2017
Specification Definition/ Planning
Hardware Design/ FPGA/ Board/
Firmware
Software Design/ System
Integration/ Evaluation/
Test
Manufacturing
Cloud Services
(SaaS, Public, etc.)
17
Agenda
• Role of Edge Computing
• SoC FPGA Solution for Edge Computing
• Edge Computing GW and IoT Solution
• Summary and Next Steps
18
Summary and Next Steps
• Edge Computing accelerates response-critical IoT market
• SoC FPGA Edge Computing GW enables real-time and customizable configuration usage
• Vision: Genetic IoT solution through Deep Learning and SoC FPGA
• FUJISOFT* is the IoT integrator from design to Cloud
Next Step:
Accelerate SoC FPGA solution in IoT segment with FUJISOFT.
Please contact me ([email protected]) .