Upload
lamthien
View
216
Download
3
Embed Size (px)
Citation preview
Challenges in Wireless and Mobile Network Applications
Jie Wu
Dept. of Computer and Information Sciences
Temple University
Overview
1. Introduction Professional Activities
Current State: Wireless and Mobile
2. Sustainable Networking Green Networking
3. Mobility: Foe or Friend Foe: Recovery and Tolerant Schemes
Friend: Mobility Control
4. A Bigger Picture Applications in Cloud and TeraGrid
5. Future of Networking US GENI, SEES, and IGNITE
China Internet of Things
1. Introduction: Professional Activities
MANETs/Sensor Nets
Edtior: IEEE TMC and TC
General Chair: MASS and DCOSS
Program Chair: INFOCOM and MASS
Panel Chair: INFOCOM and MobiCom
Committee: INFOCOM, MobiHoc, and ICNP
Distributed Systems
Chair: IEEE TC on Distributed Processing (TCDP)
Executive Program Vice Chair: ICDCS
Committee: ICDCS, HPCA, and SRDS
Parallel Processing
Former Editor: IEEE TPDS and JPDC
General Chair: IPDPS
Committee: IPDPS and ICPP
Center for Networked Computing
Center for Networked Computing
Networks: wireless and mobile
Jie Wu, Center Director
Computer and Information Sciences, Temple University
Mobile Infostation
Roadway Sensors
Opportunistic
High-Speed Link
(MB/s)
Network security and privacy
Social networks and cloud computing
• Internet (NSF GENI)
• Mesh networks (NSF MRI)
• Sensor networks (NSF NeTS)
• Delay tolerant networks (NSF TC)
• Underwater networks (Navy Yard)
• Vehicular networks (SEPTA Regional Rail)
• Wireless networks (ARO)
• RFID and medical applications (PA)
• Online social networks (Amazon & Microsoft))
High performance computing
• NSF GPU/CPU supercomputer (MRI)
2. Current State: Wireless and Mobile
Current Different types: PDA, BlackBerry,
Laptop
Internet connections: more and more wireless
Node mobility
(Near) future 1 billion vehicles
5 billion RFID
10-15 billion sensor/embedded devices
Future: anytime, anywhere
(a) Edge of the Internet
(b) General way of data transmission
3. Sustainable Networking Green networking
Minimization of the Green-House Gases (GHG)
Opposite: over-provisioning and redundancy
Possible solutions
Selective connectedness
Turn off unused resources at the edge of the networks: connected dominating set (CDS)
Virtualization
Allow more than one service on one piece of hardware
Proportional computing
Consume energy proportional to utilization level: adjustable transmission range through topology control
Green Networking: A Wireless Sample
Direction antenna Energy-efficiency
Interference reduction
7
4. Mobility: Friend or Foe
Node mobility is considered to be undesirable in MANETs using a connection-based model
Recovers from and tolerates “bad” effects caused by mobility
Nodes are assumed to be relatively stable
Foe: Recovery and Tolerant Schemes
Recovery Scheme
If a routing path is disrupted by node mobility, it can be repaired quickly
E.g., route discovery and route repair
Tolerant Scheme
Masks the bad effects caused by node mobility
E.g., transmission buffer zone and view consistency
Mobility as a Serious Threat
Mobility threatens localized protocols that use
local information to achieve certain global objectives
“Bad” decisions occur because of Asynchronous sampling of local information
Delays at various stages of handshake
Mobile node movement
Local Information
1-hop information
2-hop information
3-hop information
k-hop information
Discovered via k (1, 2 or 3) rounds of Hello exchanges
Time-space view
Snapshot: a global state in time-space view
Hello interval time
Applications
• Energy saving:
– Sleep mode
• Connected dominating set
(CDS)
• Wu and Li’s 2-hop
neighborhood solution
– Adjustable transmission
range
• Topology control (TC)
• Li, Hou, Sha’s 1-hop location
solution
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
Virtual backbone (CDS)
Topology control (TC)
Two Technical Issues
Link Availability
How protocols deal with imprecise neighborhood information caused by node mobility and delays
Inconsistent Local Views
How each node collects and uses local information in a consistent way
w x w
y
x w x
y
y
Tolerant Scheme I (link availability)
A buffer zone is used in existing protocols without having to redesign them.
Tolerant Scheme II (inconsistent local view)
Consistent Local View
Each view keeps a version by using a timestamp
Conservative Local View
Maintaining a window of multiple views
New-view(i)= F(view(i), view(i-1), …view(i-k))
where F: {union, max, min, …}
(More information on tolerant schemes: Wu and Dai, IEEE IPDPS 2004,
IEEE INFOCOM 2004, IEEE TMC 2005, IEEE TPDS 2006)
Sample I (inconsistent local view)
Wu and Li’s marking process (for CDS
construction) Node u is marked if there are two unconnected neighbors
Node u is unmarked if its neighbor set is covered by several connected marked nodes with higher IDs
Friend: Mobility Control
Movement-Assisted Routing Views node movement as a desirable feature
Store
Carry
Forward
Random Movement
Epidemic Routing (Vahdat & Becker 00)
• Nodes store data and exchange them when they
meet
• Data is replicated throughout the network
through a random talk
Controlled Mobility
Message Ferrying (Zhao & Ammar 03)
Special nodes (ferries) have completely predictable
routes through the geographic area
Mobility-Assisted Routing
– Replication
• Single copy vs. multiple copy
• E.g., spray-and-wait and spray-and-focus
– Knowledge
• Global vs. local information
• Deterministic vs. probabilistic information
• E.g., MaxProp
(Predict-and-relay: Quan, Cardei, and Wu,
ACM MobiHoc 2009)
Routing in a Cyclic MobiSpace
– Challenges
• How to perform efficient routing in
probabilistic time-space graphs
– Definition (ti,p)
• p is the contact probability of two nodes in ti .
2008-5-29 ACM MobiHoc 2008
Probabilistic Time-Space Graph
A common motion cycle T (=60)
2008-5-29
Probabilistic state-space graph Remove time dimension
Links are labeled: d / pmax (delay/max transition
probability)
2008-5-29
Iterative Process
Iterative steps
Step t+1 based on step t
Ordering of neighbors
pi ≤ pimax and i pi = 1
vst+1 minp1, p2, p3… {p1(d1+ vs1
t)+
p2(d2+ vs2
t)+ p3(d3+ vs3
t)+…}
2008-5-29
Expected Minimum Delay (EMD)
Using EMD as the delivery probability metrics
Optimal single-copy forwarding: Liu and Wu MobiHoc 2008
Optimal prob. forwarding with hop constraints
Single copy: Liu and Wu MobiHoc 2009
Multiple copy: Liu and Wu MASS 2009
Other Challenges
• Intermittent connectivity
– Node mobility
– Unstable wireless links
– Scheduled on/off sensor nodes
• Mobility
• Connectivity
• Complexity
• Bandwidth/storage
• Latency
• Security
• Multicast/broadcast
(Wu and Wang, MASS 2010)
B
D
Connectivity
• (u,v) - connectivity under time-space view
– All i, (u(i), v(i))
– Exist i, (u(i), v(i))
– All i, j, (u(i), v(j))
– Exist i, j, (u(i), v(j))
View(i) View(i+1) View(j)
View window Time
Space
u v
Evolving Graph and Extensions
Time sequence: t1, t2, ..., tL
Gi = (Vi, Ei): subgraph in time [ti, ti + Δ]
Evolving graph
(V, E), where (u,v) = {i | (u, v) є Ei}
(i is called a label)
Weighted evolving graph
(u, v) = {(i, wi) | (u, v) є Ei}
where weight wi can be
bandwidth, reliability, and latency
Optimization Problems Optimization
Earliest-completion
Fastest
NSF grant: A New Algorithm an Graph Model for Networking in Challenged Environments, 2009
Minimum-hop
Maximum-bandwidth
Maximum-reliability
J. Wu, Graph, Combinatorics and
Computing, 2010
Solution: slicing and virtulation
Complexity
Managing complexity of time-space graphs
Lossless translation method
Time-space to state-space (state explosion issue)
Lossy comprehension method
Removing time using averaging in hierarchical
routing
E.g. contact information compression
(Liu & Wu: Scalable Routing in Delay Tolerant Networks,
ACM MobiHoc 2007)
Opportunities
Increasing system performance
Routing capability
Network capacity
Security
Sensor coverage
Information dissemination (mobile pub/sub)
Reducing uncertainty in reputation systems
(Li and Wu, IEEE INFOCOM 2007)
5. Future of Networking Data Management In-network processing
Tradeoffs among communication, computation, and storage
Theory Rigorous model and scaling
properties
Swarm intelligence
Social Networks Small-world (six degrees of
separation)
Scale-free networks (power-law)
Multi-disciplinary
Network Science: Hype or Reality?
Moderator at ACM MobiCom’09
Jie Wu (Temple U., USA)
Panelists
Anthony Ephremides (U. of Maryland, College Park, USA)
Chuanxiong Guo (Microsoft Research-Asia, China)
Peter Steenkiste (Carnegie Mellon U., USA)
Taieb Znati (NSF, USA)
A Bigger Picture: Cloud & TeraGrid Cloud computing: software, platform, infrastructure
Wireless and mobile applications in cloud computing
Commercial
cloud services
IBM Blue Cloud Microsoft's
Windows Azure
Amazon Web
Services
Amazon S3
Amazon EC2
......
Google Google AppEngine
Google Calendar
…
TeraGrid User Community
Blue: 10 or more PI’s
Red: 5-9 PI’s
Yellow: 2-4 PI’s
Green: 1 PI
1000 projects, 4000 users
Wireless @Temple University Metropolitan WiMax deployment
Part of GENI project
Joint project with Drexel University and City of Philadelphia
Our goal: Provide wireless coverage for downtown Philadelphia
Proposed WiMAX coverage. Downtown Philly + 60,000 students (Temple and Drexel)
Sensor-based Telemedicine through VPNs
Part of US IGNITE initiatives (NSF and OSTP)
AWS education grant
• Improve healthcare via continuous monitoring at home
• Sensors collect physiological data and relay to smartphone
• Data securely archived in public cloud for easy access
• Data processing and mining using supercomputer facilities
• Tied together using L2VPN to ensure QoS and security
.
Microsoft Hawaii
NSF MRI CPU/GPU Supercomputer
• Provide access to researchers through GENI framework.
Research Projects Tourist applications using city-wide WiMax
WiMax centric bodynet for telemedicine
Content delivery networks using WiMax
WiMax mobile surveillance for law enforcement
WiMax enhanced mobile urban sensing
US NSF: SEES Focus Areas
Research at energy-environment-society nexus
Novel energy production, harvesting, storage, transmission, & distribution
Adoption, socioeconomic, and policy issues related to adaptation & mitigation strategies
Understand/optimize life-cycle energy & carbon footprint of natural, social & built systems
Societal factors: Vulnerability, resilience &
sensitivity to regional environment change
What is Internet-of-Things (IOT)
Internet connects all people, so it is called “the Internet of People”
IOT connects all things, so it is called “the Internet of Things”
IOT is a communication network connecting things which have naming, sensing and processing abilities
感知中国 Sensing China
有名有姓 长有五官 大脑思考 语言交流
Naming: name Sensing: 5 sense organs Processing: brain Communication: language
China is Acting
NSFC will focus on the following IT topics:
• Post Moore era
• Nano Technology
• Robotics
• WSN (Sensing China)
• Beyong 4G
• Data Mining in Complex Networks
• Cloud Computing
• Cyber Space and Security
• Complex Networks and Control
• Social Computing
Questions
Why “Sensing China”
Chronicle Reason -天时 “World of Science”
China missed the chance many times
1760-1800: Industrial revolution in Europe.
1860-1900: Westernization Movement.
Early 20th century: Outside intrusion
Late 20th Century: Inside chaos - Cultural revolution
Geographical Reason - 地利 Other countries
Japan has U-Japan,2004
Korea has U-Korea,2004
US has Smart Planet,2008
China is becoming the 2nd largest economy
China should have its own strategy to continue its economical growth.