1
Toward a Framework for Power Control in Cellular
Systems

• By Zvi Rosberg and Jens Zander
• EE 360
• May 16, 2001
• Timothy J. Peters

2
Outline

• Purpose of power control
• Dominant factors affecting power control
• Measurable information
• Algorithm requirements
• Public knowledge bank
• Analysis and Conclusion

3
Purpose of Power Control

• Alleviate co-channel and cross-channel
  interference to increase resource sharing
• Resources are generally shared using orthogonal
  or semi-orthogonal techniques

4
Purpose of Power Control

• Orthogonal techniques are not perfect
  Interference results
• Causes of interference
• FDMA (Doppler shift, imperfect technology)
• TDMA (Imperfect sync, multipath, incorrect time
  slots)
• Spread Spectrum (Multipath, near-far, imperfect
  sync)

5
Dominant Factors Affecting Power Control

• Signaling, modulation scheme, and multiple access
  method
• Link orientation (uplink or downlink)
• Environment morphology and topology
• Speed of mobile terminals
• Cell hierarchy
• Connection type (Continuous or discontinuous)

6
Dominant Factors Affecting Power Control

• Signaling, modulation scheme, and multiple access
  method
• Example
• Narrowband TDMA signal (e.g. GSM) will have few
  interfering signals
• CDMA signal (e.g. IS-95) will have many
  interfering signals
• RESULT Gaussian statistics may be applicable
  for the CDMA signal, making power control easier

7
Dominant Factors Affecting Power Control

• Link orientation (uplink or downlink)
• Example
• Base stations are usually stationary and
  elevated.
• Mobile units are moving and are located among
  many obstacles.
• Asymmetry of broadcast and MAC channel
• RESULT Uplink and downlink experience different
  shadowing and multipath which affects power
  control.

8
Dominant Factors Affecting Power Control

• Environment morphology and topology
• Example
• Outdoor rural environments are dominated by
  line-of-sight, with large obstacles at great
  distances.
• Outdoor urban environments are influenced by
  reflections from corners and line-of-sight.
• Indoor environments are affected by shadowing
  from walls
• RESULT Different environments have different
  interference statistics, which affects power
  control.

9
Dominant Factors Affecting Power Control

• Speed of mobile terminals
• Example
• With fast-moving mobiles, shadowing may not be a
  problem.
• Slow-moving or stationary mobiles will be
  affected by shadowing.
• RESULT Interference statistics are dependent on
  the speed of the mobile, which affects power
  control.
• Broadband Communications on the Highways of
  Tomorrow

10
Dominant Factors Affecting Power Control

• Cell hierarchy
• Example
• A system may contain macro cells composed of
  micro cells composed of pico cells.
• Transmission is constrained by the different
  power limits and resources available within each
  cell.
• RESULT Interference will be asymmetric,
  resulting in a different statistical model.

11
Dominant Factors Affecting Power Control

• Connection type (Continuous or discontinuous)
• Example
• Voice traffic is continuous and delay
  constrained.
• Data (packet switched) is bursty and somewhat
  delay constrained.
• RESULT Interference is affected by the nature
  of the traffic.

12
Measurable Information

• For practicality, power control algorithm should
  be distributed and use only local information
• Measurements and power updates must be fast
  enough
• Subject to measurement errors, sampling errors,
  and information aging

13
Algorithm Requirements

• Fixed or variable increments
• Fixed or variable time steps
• Distributed and asynchronous updates
• Stability of the control process
• Ease of implementation and robustness

14
Public Knowledge Bank

• Power control and many other parameters require
  full system simulation
• Each research group creates its own unique
  simulation scenario
• No benchmark scenario exists
• Authors propose an open and public knowledge bank
  to allow researchers to compare their models

15
Analysis and Conclusion

• Power control is a complex problem, affected by
  many factors
• System simulations have become complicated and
  time consuming
• Knowledge Bank will not work due to code
  compatibility issues, profit takers, and egos,
  among other reasons