Transcript Slides: Challenges in Mobile Networking

Research Challenges in
Mobile Networking and
Computing
Thomas F. La Porta
Professor, Department of Computer Science and Engineering
Penn State University
Outline
• Overview of Mobile Networking and Computing
• Mature Research
– Micro-mobility
– System interworking
• Open Research Topics
– Security proposal
• Summary
0
Mobile Internet - Ubiquitous Communications and Computing
Wireless access to networks will soon be the most popular method
• Today: 850M phone lines, 700M wireless subscribers
• Wireless growth far exceeds wired growth
Internet is driving network growth
33
Global Access Lines (billion)
• Internet users still doubling yearly
Ubiquitous wireless data has stalled…
• low data rates
• limited terminals
22
wireless
Wireless
Cable
cable
Wireline
wireline
11
• limited applications
…Mobile computing has not
• lap tops are everywhere
0
0
1958
1958
1978
1978
2005
1998
20
• dial-up service is pervasive
 Hindered by low performance, poor connectivity, and inconvenience
1
Mobile Communications Today: Tale of 2 Networks
Cellular Telecommunications Network
Radio
Telephone
Network
Wireless
Controllers
Wireless
Gateways
Internet
• Network tailored for voice – very low bandwidth
• Devices not suitable for Internet and computing applications
 Despite high penetration & coverage, Internet access has fizzled
Telephone
Network
The Internet – Wireless Enterprise Networks
Radio
Edge
Router
Access
Router
Internet
• Network tailored for best-effort data traffic – high bandwidth, no controls
• Supports general computing and data networking applications
 Gaining high density in hot-spots, but no ubiquitous coverage
2
Evolution to Multiservice networks
3G Cellular
Networks
• Outdoor Areas
• High Mobility
Radio
Controller
Aggregation
Router
Access
Router
Presence
Enterprise
Location
Networks
Access
Router
• 802.11++
Core Internet
• Local Mobility
Backbone
• Packet Voice
Aggregation
• High Data Rates
Authentication
Router
4G Air
Interface
4G
4G
Radios
Radios
Access
Router
Urban
Networks
• Broadband
Distribution
Networks
• High Speed
Pico Cells
Aggregation
Router
Ad Hoc • Allow People to
network
Networks
• Self Configuring
Home
Networks
• DSL/Cable
• High Speed
Internet Access
• End-to-end Internet
– common mobility management and control
– common transport infrastructure
– common services infrastructure
 Unifies various technologies (wireless, wireline, mobility)
3
Examples of Individual Multiservice Networks
Wireless Networks
• 1980’s – voice services
• 1990’s – voice services with very limited data capabilities (e.g., short messaging)
• 2000’s – 3G multimedia networks (voice, data, multimedia)
Internet
• Original use – best effort data (email, remote login, file transfers)
• 1990s – some multimedia (video and audio streaming, conferencing)
• 2000’s – merging of data, voice, and multimedia (Internet telephony in enterprises)
Next trend (already ongoing): merging individual networks
• Impacts:
– transport
– control protocols
– services infrastructure
– management
– security
4
Merging of Networks
Many challenges due to diverse applications, legacy systems, and philosophies
• Voice (aka Public Switched Telephone Network) and Data (aka The Internet)
Characteristic Classic Voice
Switching
Circuit (all
reserved)
End Devices Dumb
Classic Data
Packet
(queuing)
Smart
Intelligence
Network-based
Services
For connections,
billing
Client/Hostbased
Applications
Comments
Classic, old
problem
Mixed with
wireless
Starting to
change
Starting to
change
• Public and Private (corporate networks)
Characteristic Public
Management End-to-end
Security
Left to end
systems
Performance Best effort
Billing
Private
Based on need
Used to be
closed
Typically, good
None, simple
Comments
Coordination
Big problems,
+ wireless!
Need to share
Gaurantees?
• Interworking is a big problem
– service definitions, protocols, profiles, …
End result: Telephone network will get dumber, the Internet will get smarter
5
Where are we?
Seamless high-speed ubiquitous network access
• new access technologies – many already exist
• mobility between networks – active research area ()
Internetworking
• drive towards common control paradigm – controversial ()
• interworking between networks – islands exist ()
• performance – new demands on network controllers
Security
• eCommerce, enterprise data, distributed computing – no one feels safe
• authorization and accounting – bringing order to chaos
Intelligence and Applications
• distribution between network and end devices – solutions have been vertical
6
Where are we?
Seamless high-speed ubiquitous network access
• new access technologies
• mobility between networks
– real-time mobility – Handoffs
– real-time location – Paging
Internetworking
Security
Intelligence and Applications
7
Current Internet-Based Mobile Packet Networks: Mobile IP
Permanent IP Address
Temporary IP Address
Regular routing
Home
Agent
MD
Radio
Access
Networks
Foreign
Agent
Internet
Host
Intranet
Foreign
Agent
IP Tunneling
• Handoffs always managed by Home Agent
– high update overhead
– slow handoffs
• 3G UMTS Network structure is similar
– uses specially defined (non-Internet) protocols for mobility
8
Real-Time Mobility: Domain-based Mobility
Internet
Domain
Router
R
R
Domain
Router
R
R
R
R
R
R
R
R
R
R
MD
Local mobility
Local mobility
• Distributed control: Enhanced reliability, better scaling
• Localized mobility management: Lower overhead, fast handoffs
• Based on IP protocols: Independent of access network
 Mobility
across access networks
9
Real-Time Mobility: HAWAII for handoffs
Internet
Domain
Router
R
R
Domain
Router
R
R
R
R
R
R
R
R
R
R
MD
Local mobility
Mobile IP
Local mobility
• Distributed control: Enhanced reliability
– recovery of lost state via Internet routing protocols
• Localized mobility management: Lower overhead, fast handoffs
– updates only reach routers affected by a movement
• Minimized or Eliminated Tunneling: efficient routing, easy QoS
– dynamic, public address assignment to mobile devices
 Compatible
with Mobile IP
10
HAWAII Overhead
Domain
Router/HA
Router 1
... Router 7
BS1 ... BS-20 BS1 BS20
Message
HAWA
II DR
Msg/sec
Hawaii Registration 127.8
Hawaii Updates
51.3
M-IP Registration
48.4
M-IP Updates
12.7
Total
240.2
BD
RD
r
v
LB
TR
Y
TM
g
Mobile
IP HA
Msg/sec
0
0
574
127.4
701.4
Base stations/domain router
2nd level routers/DR
User density
User velocity
Base station perimeter
HAWAII refresh timer
# of updates/message
Mobile IP lifetime
% users outside home domain
rn
MIP =
L B
B
r
p
2
L B
16 T
+
D
140
7
2
39 Km
112 Km/hr
10.5 Km
30 s
25
300 s
0.1
B
D
M
H
+
rn
=
L
B
B
+
D
p
rn
D
p
L B
16 T
L BD
B
D
M-IP
M
L B R
B
gr
2
D
r
+
2
B
16YTR
Hawaii
11
HAWAII Handoff Delay: Packet Loss
Audio (160B/20msec)
12
Real-Time Location: IP Paging
• Motivation
– dormant mode increases battery life considerably
– mobile device must be located (quickly) to deliver data
– currently each network type has unique paging protocols
• Challenges
– limit overhead (messages) while maintaining low latency
– do not impact fast-path packet forwarding
– devise scaleable solution in terms of network size and administrative domains
– support various paging algorithms
• Basic Solution
– compatible with and leveraging IP protocols
Determined by
– distributed control for scale and scope
Registration
Determined by
Paging
MD
Serving
R
Router/Switch
Home
Router/Switch
Network
Determined by
address
Host
R
R
13
Unified Paging
HA initiates page
Uniform mobility management
– wireless LANs, outdoor
Home
Agent
Internet
R
R
Old FA initiates page
Foreign
Agent
Mobile IP Foreign
Paging Area Agent
R
Domain
Paging
Area
Foreign
Agent
Any router initiates page
3 Options:
• Home Agent Paging
– Home agent buffers packets and initiates page to all Foreign Agents
– Can be controlled by corporate network
– Does not scale
• Foreign Agent Paging
– Last active Foreign agent buffers packets and initiate paging
– Distributes load
• Domain Paging
– Fully distributed, very scaleable and reliable
14
IP Paging: Domain Paging
Any router
initiates paging
Internet
Domain
Router
IP
Paging Area R
R
Domain
Router
R
R
R
R
R
R
R
R
R
R
• Basic Procedures
– dormant MH does not send updates unless crossing Paging Area
– packets sent to last known domain
– any router in old path to Mobile Host may initiate paging
• Characteristics
– more complex: changes to routers in access network
– excellent scaling: load distributed among all routers
– excellent scoping: only routers in a paging area need topology information
– excellent reliability: failed routers are routed around (OSPF)
15
IP Paging: Latency Results
• Latency (from Stanford University Local Area traces)
– Foreign Agent and Domain Paging scale best: distributed load
– Domain paging slightly more scaleable (~10%): dynamic load balancing
• Messaging Overhead
– Domain Paging has lowest overhead: no paging related updates to Home Agent
16
Where are we?
Seamless high-speed ubiquitous network access
Internetworking
• drive towards common control paradigm
– Internet Control
• interworking between networks
– support for legacy systems (and stubborn operators!)
• performance
– efficiency
Security
Intelligence and Applications
17
Internet Control and Interworking: Unified Mobility Manager
Provide Home Location Register and Internet functionality
Current
User/mobile location management
HLR
Cellular
Integrate
HLR/VoIP server
functions
Security control (authentication
center, IP security server)
Service profile DB management
VoIP server
IP security
server
Internet (VoIP)
– mobility management
– security
– profile management
• Support multiple interfaces
– Wireless: ANSI 41, GSM / UMTS MAP
– Internet: VoIP (SIP, H.323), AAA (RADIUS)
Integrated
User
Profile
Database
Security
manager
MAP
/ SS7
Protocol
Gateways
• Integrate 3 key functions of HLR & IP (VoIP) servers
User
location
manager
VoIP / IP
AAA / IP
Cellular
Internet
ANSI41
/ SS7
Cellular
18
Unified Mobility Manager: Architecture
Core servers
- Common functions across protocols
- Interface via abstraction of basic services
- Multiple server instances on a cluster of processors for system scalability
- Interworking function between protocols: O(n) versus O(n2) complexity
Protocol gateways
- Perform protocol specific controls
- Facilitate new protocol introduction
Integrated user profile DB
- Common database for a ‘user’
CDMA 2000
GSM/UMTS
PG
PG
Wireless Data
PG
VoIP Servers
PG
Core
Core
Operations Servers
Integrated
user
profile
19
UMM – Location Management Performance
• 2 Basic Scenarios Considered: UMTS Network with HLR or UMM
• 2 call scenarios – goal: calls travel over IP network
– call from PSTN to UMTS phone (PSTN to cellular)
– call from SIP phone to UMTS phone (Internet to cellular)
PSTN call
HLR/
UMM
HLR & UMM are the same
GW
Switch
Serving
Switch
PSTN
BS
SIP Call
SIP
Internet
MGW
HLR
GW
Switch
PSTN
Serving
Switch
BS
SIP-HLR
• uses PSTN for transport
• inefficient
UMM
Internet
MGW
Serving
Switch
BS
UMM
Internet
Serving
Switch
BS
SIP-UMM
• uses packet for transport
• efficient
SIP-UMM (SIP e2e)
• most efficient
20
UMM Analytical Results
• Migration to packet
– short term: performance degradation from circuit/packet translation
– long term: packet end-to-end improves performance
21
Research Topics: Intelligence For Network Transport
Balance between simple networks and functional networks
Ad Hoc Networks
– self-configuration
– very dynamic routing
– Challenges: security, performance, relay networks to enhance cellular telecom
Sensor Networks
– power-aware component to most functions (routing, etc.)
– nodes participate in processing information
– Challenges: placement, distribution of functions
Traditional Networks
– multiclass admission control & charging algorithms
– IP Services platforms (firewalls, web redirection, …)
– Challenges: heterogeneous networks, network architecture
Heterogeneous Networks
– real-time mobility between networks
– Challenge: vastly different network characteristics
22
Research Topics: Intelligence for Services
Support for creativity, rapid deployment, and wide accessibility
Network architectures
– separation of transport and service control
– re-use across network types
– Challenges: scalability, performance, reliability
Protocols
– interworking across systems
Services
– legacy systems
– Challenges: correct semantics, functionality mapping, software
Heterogeneous networks
– access from different networks
– access via different protocols
– Challenges: security, flexibility, performance
Transport
23
Research Topics: Security
Performance/Security Tradeoff
Traditional Wireless
• 3G data services require network-based enhancements
• Challenges – limiting performance impact, or improving performance
Ad hoc wireless, sensor networks
• existing solutions are not scalable
• Challenges – trade-offs of security and performance
General
• end-to-end security solutions, including network, devices, OS, …
24
Performance Focused Security
Goal: Allow network-based performance enhancements in a secure environment
• Specific Problem: Data performance in wireless networks is poor
– network-based solutions have been proposed for 3G
– caching has been proposed for ad hoc and relay networks
– proposed solutions contradict security models for mobile VPNs
• Sketch of solution
– allow at least one intermediate “router”, positioned at the edge of the network, to view and
process a portion of a secure packet
– trade-off level of security for performance
• Benefits
– higher performance data services
– access to network-based value added services with limited security trade-off
25
Security Solution for 3G: Mobile Multi-layered IPSec
Client
Wireless
Router
Network
End-to-end encryption for payload
Corporate
Network
Firewall
}
Host
Example
Concatenated encryption for control information
• Define multiple zones in a packet (as in ML-IPsec by Zhang and Singh)
– each zone has an associated security association (keys, etc.)
– each zone may be terminated in one or more places inside the network
• Research: Initialization, Mobility, Performance
– distribution of keys and initialization
– maintenance of security during mobility
– performance impact
– implementation issues
26
Ubiquitous Mobile Computing - Summary
The potential still remains…
…Hard work required – networking: end-to-end systems issues with access and core
– intelligence: distribution between device and network (dynamic & flexible)
– applications and support services: retain generality
– new network architectures
… with Imagination – new ways to use mobile capabilities
– perhaps extension of wireline network paradigms is wrong!!!
27