Title: Understanding the Effects of Hotspots in Wireless Cellular Networks
1Understanding the Effects of Hotspots in Wireless
Cellular Networks
- Infocom 2004
- Speaker Bo-Chun Wang
- 2004.7.23
2Outline
- Introduction
- Related Work Motivation
- Modeling Hotspots
- Experimental Results
- A Fluid Flow Model for Modeling Cellular Networks
- Analytical Modeling of Hotspots
- Conclusion
3Introduction
- Hotspots
- occur whenever there is contention among users
for bandwidth - lead to blocked and dropped user
- impact the performance of the network
4Previous work
- Consider hotspots with reference to load
balancing and congestion control - Focus on algorithms and techniques to improve
capacity and performance - Lack of research that studies properties of
hotspots in detail
5Contributions
- Classify hotspots into three different types
based on some cause - Show how the different types of hotspots differ
in their impact on performance - Propose a fluid flow model to substantiate of
experimental observations - Develop an analytical model to study hotspots
6Related Work
- Sajal K. Das, Sanjoy K. Sen, Rajeev Jayaram. A
Novel Load Balancing Scheme for the Tele-traffic
Hot Spot Problem in Cellular Networks. Wireless
Networks,1998 - Johan Karlson, Berth Eklundh. A Cellular Mobile
Telephone System with Load Sharing An
Enhancement of Directed Retry. IEEE Transactions
on Communications, 1989
7Related Work (contd.)
- Jung-Shyr Wu, Jen-Kung Chung, Chang-Chung Wen.
Hot-Spot Traffic Relief with a Tilted Antenna in
CDMA Cellular Networks. IEEE Transactions on
Vehicular Technology, 1998
8Motivation
- Most of previous works only deals with hotspots
within the context of other problems - None of them identify the specific
characteristics of hotspots - A better knowledge of hotspots can help in
designing realistic simulations which can
facilitate better network design
9Some Questions addressed in work
- How are hotspots created?
- How do they affect network performance?
- Do different types of hotspots impact performance
in different ways?
10Modeling Hotspots
- Delay based there is an accident which is
delaying all the users in that cell. - Capacity based the base station of a cell can
only support a lower number of users. - Preferential mobility based there is an event
and people are moving towards a given location.
11Simulation setup and network model
- Use a square cell for simplicity in the
implementation and in the analysis. - Two types of networks
- bounce-back network This network exhibits edge
effects. - wrap-around network This network avoids edge
effects.
12Regions in a bounce-back network
13Implementation details
- Users and cells
- homogeneous users and cells
- the capacity of each cell is 50,000 units
- the bandwidth requirements of each user is 500
units - the cell latency are exponentially distributed
- Mobility model
- random walk mobility model except in the case of
the preferential mobility hotspots
14Implementation details (contd.)
- Performance measures
- network utilization the ratio of the number of
users in the network to the number of users that
the network can support - steady state utilization a networks maximum
utilization without loss
15Number of hotspots
16Placement of hotspots
17Clustering of hotspots
18Local and global impact
19Local and global impact (contd.)
20A Fluid Flow Model for Modeling Cellular Networks
21One dimensional chain network
22Two dimensional bounce-back network
23Modeling Hotspots Using an Enhanced Fluid Flow
Model
24Modeling Hotspots Using an Enhanced Fluid Flow
Model (contd.)
25Modeling Hotspots Using an Enhanced Fluid Flow
Model (contd.)
26Analytical Modeling of Hotspots
- Consider the network to be a closed network of
M/M/B/B queues - Find the marginal probability that there are
exactly ki k jobs at the ith node in such a
closed network
27Analytical Modeling of Hotspots (contd.)
28Conclusion
- Present a detailed study of hotspots in wireless
cellular networks - Identify three types of hotspots based on
different causes and study their different
properties - Develop two analytical models to substantiate
simulation results