1. Stefan Pettersson Mid Sweden University
Capacity Comparisons of Wireless Indoor Networks with Non-uniform User Distribution
Stefan Pettersson
Mid Sweden University
Department of Information Technology and Media
Mid Sweden University

2. Department of Information Technology and Media
Outline
Indoor Model & Assumptions
Wireless System Comparisons
Centralized Channel Assignment
Distributed Channel Assignment
Fixed Channel Assignment
Numerical Results
Conclusions
Department of Information Technology and Media
Mid Sweden University

3. Department of Information Technology and Media
Indoor Scenario
A building floor with
20 base stations in every
second office
Meters
Department of Information Technology and Media
Mid Sweden University

4. Department of Information Technology and Media
User Distributions
Uniform distribution of users
on the covered building floor
Indoor ”hotspot” with five
times the user concentration
within a 20 by 20 m square
Department of Information Technology and Media
Mid Sweden University

5. Sector Antenna Pattern
Non-overlapping sector patterns
No channel reuse between sectors
Normalized 0 dB gain in the front lobe
-15 dB gain in the back lobe
Constant vertical pattern within lobes
One horizontal sector pattern
at a four-sector antenna site
Department of Information Technology and Media
Mid Sweden University

6. Centralized Channel Allocation
The mobiles measure on beacon channels and the central unit builds a link gain matrix, G.
A feasibility check is performed prior to new channel allocation to ensure the quality of all links
Least-interfered channel selection
SIR-balancing power control
Department of Information Technology and Media
Mid Sweden University

7. Distributed and Fixed Channel Allocation
Distributed DCA
No feasibility check
Least-interfered channel selection
SIR-balancing PC
Fixed ch. allocation (FCA)
No feasibility check
Random channel selection
SIR-balancing PC
Department of Information Technology and Media
Mid Sweden University

8. Department of Information Technology and Media
Numerical Analysis
Computer simulation in downlink
Orthogonal channels
No mobility
Log-normal shadow fading with 12 dB std. dev.
L [dB] = Log10(DTx-Rx)
User assignment failure, u: The fraction of users that did not get a channel or got a channel with too low quality
Relative load: user/channel/cell
Capacity: The load where u equals 0.02
Department of Information Technology and Media
Mid Sweden University

9. Uniform Distribution Omni-directional Antennas
Capacity
FCA: 0.06
Distr. DCA: 0.31
Centr. DCA: 0.54
User assignment failure
Relative load
Department of Information Technology and Media
Mid Sweden University

10. Hotspot Omni-directional Antennas
Capacity
FCA: 0.05
2 %
Distr. DCA: 0.25
18 %
Centr. DCA: 0.44
User assignment failure
Hotspot vs. no hotspot
Relative load
Department of Information Technology and Media
Mid Sweden University

11. Uniform Distribution Sector Antennas
Capacity
FCA: 0.34
Distr. DCA: 0.46
Centr. DCA: 0.65
User assignment failure
Relative load
Department of Information Technology and Media
Mid Sweden University

12. Hotspot Sector Antennas
Capacity
FCA: 0.24
380 %
Distr. DCA: 0.32
28 %
Centr. DCA: 0.45
2 %
User assignment failure
Sectoring vs. no sectoring
Relative load
Department of Information Technology and Media
Mid Sweden University

13. Department of Information Technology and Media
Conclusions
The DCA systems suffers more from non-uniform user distribution
The Centralized system outperforms other systems with both user distributions
Sector antennas improve the performance for all investigated systems
Fixed channel assignment benefit the most from sectoring
Department of Information Technology and Media
Mid Sweden University