Transcript Strategies, Behaviors and Discounting in Radio Resource

13. ComNets-Workshop 2006
Smart Caching in Mobile IP
enabled Networks
Stephan Göbbels
Chair of Communication Networks
RWTH Aachen University, Germany
13. ComNets-Workshop, Mobil- und Telekommunikation
March 31st, 2006, Aachen, Germany
Stephan Goebbels, ComNets, RWTH Aachen University
Overview
Overview – Introduction – Smart Caching – Simulation - Outlook
• Motivation and Introduction
• The Smart Caching Approach
• System Concept
• Simulation
– Scenario
– Results
• Conclusion and Outlook
Stephan Goebbels, ComNets, RWTH Aachen University
2
Motivation
Overview – Introduction – Smart Caching – Simulation - Outlook
•
Strong development towards All-IP networks
•
Integration of various wireless networks
•
Different performance characteristics of involved communication
technologies
– Heterogeneous network structure
– Combination of wired and wireless networks
– Ambient Networks
– Always Best Connected
– Intersystem handovers
–
–
–
–
Maximum and average throughput
Packet delay and latency
QoS
…
•
Goal: Improve user satisfaction and network performance
•
New upcoming services
– Video on Demand
– Online VCR
– IPTV
Stephan Goebbels, ComNets, RWTH Aachen University
3
Introduction
Overview – Introduction – Smart Caching – Simulation - Outlook
•
Aggregation of mobile
communication networks
– Integration of buffer on aggregation
level
– Closest point between IP gateways
(SGSN, gateway router, …)
•
•
Server
Backbone
IP world
Smart Caching:
– Caching Media Port (CMP)
– Decoupling of backbone and radio
subsection of the end-to-end
connection
– Optimal utilization of available
wireless bandwidth
– Virtually continuous broadband
connection (for non real-time
services)
Aggregation
Router
CMP
MAP
Aggregation
Level
WLAN
Radio
Level
GPRS
Hierarchical Mobile IP shows similar
structure to Smart Caching
– Integration of Mobility Anchor Point
(MAP) and CMP
Stephan Goebbels, ComNets, RWTH Aachen University
4
The Smart Caching Approach
Overview – Introduction – Smart Caching – Simulation - Outlook
•
Aggregation of discontinuous high bandwidth wireless links (WiFi)
and continuous basic coverage (Cellular networks – GPRS/UMTS)
•
Idea:
•
Store enough data at the end device in order to bridge idle gaps
and pretend a continuous service
– Buffer data if wireless link is overloaded
– Forward data if radio bandwidth is available on top of current traffic
 Full utilization of wireless link – maximum throughput
– Buffered video data
– Cached WWW sites
– …
•
CMP between SGSN and WLAN Gateway router
•
Integration with Mobile IP
– Micro mobility by MAP and Macro mobility by Home Agent
– MAP can trigger data delivery of CMP
– MAP switched => new CMP
Stephan Goebbels, ComNets, RWTH Aachen University
5
System Concept
Overview – Introduction – Smart Caching – Simulation - Outlook
•
Server
Caching Client
Caching Server
Sending data
Caching
Packet request
Data forward
Continuous traffic on first
hop
Ack
Packet request
…
Major part of control traffic
only on second hop
Client
Data request
…
•
– Client entity to
communicate with service
provider
– Server entity to forward
traffic to end device
…
•
Caching node (CMP):
No ack
Stop
delivery
Location update
•
Packet request
Data forward
…
•
Better performance of
communication (TCP,
ARQ,…) and session
management protocols
Location update triggers
data delivery
Stephan Goebbels, ComNets, RWTH Aachen University
6
Simulation Scenario
Overview – Introduction – Smart Caching – Simulation - Outlook
•
•
•
•
•
•
Applied service: Video streaming (400 kbit/s)
CMP and MAP are collocated in one gateway node
Heterogeneous
network structure
by only partial
coverage of 10
IP world
WLAN Access
Points (802.11b –
11 Mbit/s)
Varying coverage
d
ratio (10 – 75%)
CMP / MAP
Consummation
rate 300 kbit/s
AP1
AP2
Legacy network
technology for
reference reasons
...
0m
160m
Stephan Goebbels, ComNets, RWTH Aachen University
AP10
1440m
7
Simulation Results
Overview – Introduction – Smart Caching – Simulation - Outlook
• Currently perceived
user throughput
2500
Current Throughput [kbit/s]
Throughput
Coverage
2000
1500
• Optimization of
wireless link
utilization
1000
500
0
0
50
100
150
200
250
300
350
400
Location [m]
• Note:
– Session recovery
delay
– Full utilization of
wireless link
450
500
• Increased throughput
after first coverage
gap
Stephan Goebbels, ComNets, RWTH Aachen University
8
Simulation Results
Overview – Introduction – Smart Caching – Simulation - Outlook
• Average throughput for long-term session
• Increase of average throughput
• Comes close to the maximum reachable
throughput
• Doubles Throughput
in 50% coverage
scenario
• Effect is amplified
if coverage ratio
decreases (as long
as coverage zones
are not fully utilized)
Cumulative Throughput [kbit/s]
450
400
350
300
250
200
150
Caching-enabled
Reference
maximum Throughput
average Throughput
100
50
0
0
500
1000
1500
Location [m]
Stephan Goebbels, ComNets, RWTH Aachen University
9
Simulation Results
Overview – Introduction – Smart Caching – Simulation - Outlook
• Decreasing
performance with
sinking coverage ratio
• But: Improvement if
compared to reference
level
• Difference between
Location [m]
full utilization and
partial utilization (25% coverage is lower limit)
• Below 25% only partial improvement (buffer
overflow)
Cumulative Throughput [kbit/s]
450
400
350
300
250
200
150
100
50
0
0
500
Stephan Goebbels, ComNets, RWTH Aachen University
75%
50%
25%
10%
Reference
1000
1500
10
Simulation Results
Overview – Introduction – Smart Caching – Simulation - Outlook
• Fill level of user equipment buffer
• Decides whether an virtual continuous service can
be provided.
• In reference scenario usually not possible
• Even 30% network
coverage is almost
enough
• In 50% case an
overspending of
resources
10000
Caching 50%
Reference 50%
Caching 30%
Reference 30%
9000
Buffered Data [kByte]
8000
7000
6000
5000
4000
3000
2000
1000
0
0
100
200
300
400
500
600
700
Location [m]
Stephan Goebbels, ComNets, RWTH Aachen University
11
800
Conclusion and Outlook
Overview – Introduction – Smart Caching – Simulation - Outlook
•
•
Integration towards ALL-IP networks will cause several
problems for service provisioning
Smart Caching is suitable protocol to enhance network
performance in new heterogeneous network structures
– Support of mostly unidirectional services
– Allows prefetching of user data
•
•
•
Benefits from integration of Hierarchical Mobile IP and Smart
Caching
Virtual continuous broadband connection in patchy coverage
areas (suitable for video services like IPTV)
Future improvements:
– Stronger integration of Mobile IP and Smart Caching protocol
– Combined GPRS – WLAN scenario
– Switching of MAP/CMP
Stephan Goebbels, ComNets, RWTH Aachen University
12
Thank you for your attention !
Stephan Göbbels
[email protected]
Any questions?
Stephan Goebbels, ComNets, RWTH Aachen University
13