Personal Journey through ICT Research

advertisement
Technology Trends and Issues in
ICT Research & Development
or
“How to achieve your dream?”
Lee Kang-Won, PhD.
ICT Synergy R&D Dept.
September 2014
2
Table of Contents
•
•
•
•
•
•
Introduction – About Me
About IBM, SKT
What is good R&D?
Major trends in ICT
Possible research directions
About R&D attitude
About Me
• Joined SKT (April 2014)
• Joined IBM Research as RSM (August 2000)
• Ph.D. Computer Science, UIUC (2000)
–
–
–
–
Thesis: Multicast for Heterogeneous Packet Flows
Kookbi Scholarship from Korea Government
C. G. Gear Outstanding Graduate Student Award
KFAS Scholarship
• Military Service at ROK Air Force (1994 – 1996)
• M.S. and B.S. Computer Engineering, SNU
– Merit Scholarship
– Student President
– Magna Cum Laude
4
Technical
•
•
•
•
•
•
•
•
•
100+ publications in premier journals
and conferences
IBM Master Inventor
KSEA Engineer of the Year (2013)
ACM Distinguished Scientist &
Senior Member
IEEE Senior Member
IBM OTAA (TPC), RDA (PMAC, ITA)
Book Author on Policy Technology
(w/ D. Agrawal et al.)
Keynote speaker: IEEE Sarnoff 2012
Invited talks: UIUC, Columbia,
Samsung, SNU, KAIST, …
5
Projects Managed - ITA Program
• Summary
– Fundamental research in
Network Science
– IBM-led consortium funded
by US Army and UK MOD
– 10 Year Program (5 + 5)
• Funding Structure
– Basic research: $100M total
– Transition contracts
additional
 ITA Technical Scope
– Tech
– Tech
– Tech
– Tech
Area
Area
Area
Area
1:
2:
3:
4:
Network Theory
Security
Sensor Information Processing
Coalition Decision Making
– My Role: Technical Area Leader for Tech
Area 1 (Network Theory)
6
Projects Managed – Cloud Computing
• Funding Agency: NIST
– 3 year (2010 – 2012)
• Goal
– Develop algorithms and mechanisms for management of large
scale cloud computing systems
• Team
– IBM: ITW, BAMS, STR, SWR
– Cornell University
• Output
– Science: 15 papers published top venues (INFOCOM, PODC,
Allerton, SMPTS, AISTATS, KDD, CLOUD); several patents
– Biz Impact: Anomaly Detection (TASP GA) Predictive Analysis
(NSN demo, Streams GA)
– Standard Impact: SPEC Cloud Benchmark
7
Technology Transfer: IBM TPC
• Background
– Configuration errors are one of
the leading causes of SAN
disruptions and maintenance
costs
(w/ Almaden)
Before
• Configuration Checking
Utility for TPC v3.2
– Validates the correctness of
SAN configuration by checking
it against best practices and
policies from field practitioners
– Diagnoses across multiple
devices and hardware/software
/firmware components
• Outstanding Technical
Achievement Award, 2008
– Still available in Version 5
(4:30)
8
After
Warning: host can see
both tape and disk
Tech Transfer: Spatiotemporal Analytics
•
Motion processing: Geofencing, Hangout, Ma
p matching, Compression
•
Fast Indexing for query: New algorithm using
prefix matching; efficient for KV stores
– 10x – 1000x speed up depending on
applications
•
Full Earth operations: handles large objects (e
.g., cargo ship, satellites), any location (e.g., ar
ound poles)
•
Transferred to SPSS, G2, Informix
•
In plan for DB2, BigInsight, Streams
•
Applications: Connected cars, (1:00) Insurance,
Location-specific monitoring, Digital billboards,
etc.
9
9
About IBM
Quiz: How old is IBM?
• Hint: MS is 39 years old.
•
•
•
•
A: 65 years
B: 81 years
C: 92 years
D: 103 years
Quiz: How old is IBM?
• Hint: MS is 39 years old.
•
•
•
•
A: 65 years
B: 81 years
C: 92 years
D: 103 years (founded in 1911)
Quiz: What does IBM stand
for?
Quiz: What does IBM stand
for?
Quiz: What does SK mean?
Quiz: What does SK mean?
• 선경 (鮮京)
Quiz: What biz did SK start?
Quiz: What biz did SK start?
• Textile (선경직물, 1953)
A little more about IBM
• In 2012
–
–
–
–
–
No.
No.
No.
No.
No.
2
4
1
2
1
largest employer in the U.S.
largest in terms of market cap
company for leaders (Fortune)
most respected company
green company
IBM Research
• 12 labs worldwide
• Milestones
–
–
–
–
–
–
–
DRAM, HDD
Fractal
FORTRAN
RISC architecture
Relational DB
Deep Blue
Watson (1:00, 2:51, 6:30, 9:40)
• 5 Novel Prizes
• 4 Turing Awards
A little more about SK telecom
• First and best
LTE-A
LTE
HSPA+
WiBro
HSDPA
S-DMB
1x EV-DO WCDMA
IS-95A/B
1996
2G
2006
HSUPA
2010
2013
2011
Domestically
First LTE
Domestically
First
Data network
HSPA+
Paradigm Shift
World’s first
HSUPA for
2007
2005
1x/EV-DO
2.5G
3G
3.5G
4G
2014
300 Mbps
World’s
first
5.76Mbps
World’s first
HSDPA
with
CDMA
2000 1X
World’s firstHandset
2003
Satellite
World’s firstDMB
2000
WCDMA R4
Provide new experience to customer with high-level service
World’s first
such as high-rate data, video telephony and Global roaming
CDMA 2000
World’s first
CDMA
3CA
Ⅲ. SK Telecom R&D
Focus Areas
NETWORK
TECHNOLOGY
INFORMATION
TECHNOLOGY
-LTE/LTE-A, 5G,
-Location-based service
-Context-awareness
-Network-enabled cloud
-Big data analytics
-Internet of Things
-Security
CONVERGENCE
TECHNOLOGY
HEALTHCARE
-Storage technologies
-Quantum Crypto
-Video/audio analytics
-In-vitro Healthcare
-Personalized
healthcare
*eICIC: enhanced Inter-Cell Interference Coordination
22
What is Good R&D?
What is Good R&D?
• “If we knew what it was we were doing,
it would not be called research, would
it?”
– Albert Einstein
Pasteur’s Quadrant
What is Good R&D?
• Important Problem
– Real world issue
• Ingenious Solution
– Trade Secret or Patent
• Biz Impact
– Can make money
Major Trends in ICT
Trend 1. Network is increasingly
being dominated by data
Trend 1. Network is increasingly
being dominated by data
• Bandwidth vs. response time vs.
availability
– SNS, multimedia, search, VR, AR
• More variable, dynamic, integrated
– Real-time OSS/BSS
Trend 2. Big opportunities for big
data
Quiz: How much data generated
between 1993 – 2012?
• Hint: 5 exabytes* generated between
3000 BC – 2003
* 1 Exabyte = 10^18 bytes = 1 M Terabytes
Quiz: How much data generated
between 1993 – 2012?
• Hint: 5 exabytes* generated between
3000 BC – 2003
• Answer: 4000 exabytes
* 1 Exabyte = 10^18 bytes = 1 M Terabytes
Quiz: If we stack books containing
4000 exabytes, how high will they
be?
• Hint: Think BIG
Quiz: If we stack books containing
4000 exabytes, how high will they
be?
• Hint: Think BIG
• Answer: 80 roundtrip times between
Earth and Pluto (160 x 5.9B km)
Trend 2. Big opportunities for big
data but …
• Didn’t crack it yet
– MNOs vs. OTTs
• Privacy: Customer sentiment
– plus regulations
• Other obstacles
– End-to-end encryption
Trend 3. IoT can provide new
opportunities
• IoT is a $19 trillion opportunity – John Chambers
Trend 3. IoT can provide new
opportunities but …
• Innovation is happening elsewhere
– Nest, drones, watch, self-driving cars
• Selling “circuits” vs. solution?
• What about security and privacy?
Direction 1. Virtualization to the
rescue
• No constraints from physical assets
– NFV, SDN, Network-enabled cloud
• “Capacity breathing”
– Auto-scaling, self-scaling
– Predictive
• Dynamic
– At any granularity, at any time scale
• Point-to-point, multicast
– Based on usage
Direction 2. Mining the Data
• More than just “big data”
– Synergy between MNO & OTT?
• Mining, sieving, refining, distilling
– What is the purpose?
– What are you looking for?
– Intelligence, Insight
• New tools
– Graph DB, column store, big table, stream
processing
– Privacy-preserving, delegated computing
Businesses are “dying of thirst in an ocean of data”
46
90%
80%
20%
of the world’s data
was created in the
last two years
of the world’s data
today is
unstructured
amount of data
traditional systems
leverage today
1 in 2
83%
2.2X
business leaders
don’t have access
to data they need
of CIOs cited BI and
analytics as part of
their visionary plan
more likely that top
performers use
business analytics
How to handle Big Data?
• Maybe we need “data refinery”
Oil refinery
Data refinery
Unstruct
ured
Data
User
Intent
Spatiote
mporal
Context
Data
Structur
ed Data
Purchase
History
CDR
Billing
Network Intelligence
Raw
Data
User
Sentiment
Social
Net Data
How to ensure safe data analysis?
How to ensure safe data analysis?
• Privacy Homorphism: One of 10 Emerging
Technologies (MIT Technical Review 2012)
Direction 3. Rich IoT
• Extreme scale
• MNO has a unique peering point
– Sees traffic, location, ubiquitous
• Can provide value-added solutions to customers
– Built-in analytics, security, AI
– AaaS, SaaS, AIaaS
What is Good R&D?
• Important Problem
• Ingenious Solution
• Biz Impact
What is Good R&D?
•
•
•
•
Important Problem
Ingenious Solution
Biz Impact
Solves the Problem Completely
– Google vs. Alta Vista, Lycos, Bing, etc.
– Facebook vs. MySpace, Cyworld, etc.
– Amazon vs. Best Buy, Walmart, etc.
Wait, this is overwhelming…
About R&D Attitude (1)
• Develop core strength; buzzwords come
and go
Buzzwords come; buzzwords go
• Big data
– data mining, parallel computing, dist.
computing
• IoT
– device, processor, sensor, communications
• Cloud
– system software, virtualization, optimization,
networks
• Cognitive computing
– AI, machine learning, neural network, NLP
About R&D Attitude (2)
• Put enough time, effort, and attention
About R&D Attitude (2)
• Put enough time, effort, and attention
About R&D Attitude (3)
• Execution is the key.
About R&D Attitude (4)
• Don’t give up so easily
Brian Acton: Co-Founder of WhatsApp
So, what is your dream?
Want to hear your thoughts.
Thank you!
Download