Radio Resource Management in Wireless Mobile Networks

1
Radio Resource Management in Wireless Mobile
Networks

• Emre A. Yavuz, Ph.D. candidate
• Supervised by Dr. Victor C.M. Leung
  Communications Lab., Elec. Comp. Eng.
• University of British Columbia, UBC
  emrey_at_ece.ubc.ca

2
Motivation

• To build mathematical models for Radio Resource
  Management and to simulate the resource
  allocation process
• in order to look for optimum algorithms and
• develop better admission and congestion control
  procedures.
• Extra capacity that is provided to the network
  will result in higher end-user average bit rates,
  low delays and BER and lower blocking/dropping
  ratios.
• Radio Resource Management will be the major
  differentiator between the overall QoS
  provisioning offered by different operators
  networks.

3
Agenda

• What makes Wireless Mobile Networks different ?
• The Radio Resource Management Methods
• Resource Usage in CDMA Networks
• Admission Control Its Logical Dependencies
• Congestion Control Actions to be Taken
• Traffic Types, Adaptation of Quality Indicators
  Cost
• Resource Management based on Utility Function
  Approach
• Pricing Frameworks
• Future Research

4
What Makes Wireless Mobile Networks Different ?

• The probabilistic behaviour of the wireless
  channel (shadow fading, rayleigh fading, path
  loss) as opposed to almost deterministic behavior
  of the wired lines.
• Soft Capacity (changes with other own cell
  interference)
• User - mobility (handovers, shadowing effects
  etc.)
• Call Dropping Probability constraint besides Call
  Blocking Probability constraint.
• Cell Coverage guarantee for each service.

5
The Radio Resource Management Methods

• Admission Control
• Handles new incoming traffic (new connections,
  handover decisions and bearer modifications)
• Congestion Control
• Manages the system when the load exceeds
  threshold
• Traffic Scheduling (mostly for non-real time
  traffic)
• Handles packet data users to initiate the packet
  transmissions and guarantee QoS through bit rate,
  BER and delay adjustments
• Power Control - to maintain radio link quality

6
Resource Usage in CDMA

• Uplink (Interference Limited)
• Downlink (Power Limited)

7
Admission Control Its Logical Dependencies

• Estimate the load and fills the system up to the
  limit without having impact on coverage and
  quality of existing connections.
• Separate admission for UL and DL.
• Uses load info from Congestion Control, Traffic
  Scheduling, Power Control and Handover Control.
• Derives the transmission bit rate, target BER,
  processing gain, initial link quality parameters.
• Initiates the forced call release and
  interfrequency or intersystem handover.

8
Congestion Control Actions to be Taken

• Optimize the capacity of a cell and prevent
  overloading by measuring application parameters
  from planning and UL DL interference.
• Congestion Control takes care of the network to
  prevent overloading and to preserve the
  stability.
• Consider load control actions on the network
  traffic.
• Lower bit rates of the rate-adaptive traffic.
• Lower SIR target based on the type of
  application.
• Interact with TS and throttle back packet data
  traffic.
• Force interfrequency or intersystem handover.
• Drop calls in a controlled manner.

9
Traffic Types, Adaptation of Quality Indicators
Load

• Quality of Service (QoS) Classes (Conversational,
  Streaming, Interactive and Background)
• Main concern Real-time and Non-real-time
  Traffic. (More compatible with scheduling when
  compared toreal-time traffic)
• Regrouping the traffic
• Adaptive rate and adaptive signal quality
  requirements.
• Adaptive rate and fixed signal requirements.
• Fixed rate and adaptive signal requirements.
• Fixed rate and fixed signal requirements.

10
Resource Management based on Utility Function
Approach

• Objective To maximize aggregate utility subject
  to
• Available transmission power and spreading codes
  (at the base station)
• Allowed interference (at the mobile stations)
• A solution based on a pricing framework, where
  prices per unit power, code or load are announced
  by the network (RNC) in order
• the users to maximize their net utilities while
• the system tries to maximize the total sum of
  utilities and
• the network tries to maximize its revenue.

11
Three Cases of Optimization

• From User point of view
• From System point of view
• From Network point of view

where ? is the price that is announced by RNC
MAX
where ki is a coefficient chosen for QoS or
price-based class
12
How should a Utility Function for a SIR Adaptive
Application be ?
U(SIRi)1/(1exp(-(a(SIRi -b))))
13
Cost Calculation based on Pricing Frameworks

• Cost can be based on a shadow price and a
  constraint related load parameters like rate,
  BER, power etc.
• Based on the rate and BER of the application
• Cost Shadow price rate SIR
• Based on the application power
• Cost Shadow price power
• Shadow price will be announced by RNC for each
  period and be adjusted dynamically based on the
  network load.

14
How does the Derivative of the Utility Function
Look Like ?
15
How does the Net Utility Function Look Like ?
16
How does the Radio Network Control make the
Upgrade/Downgrade Decision ?
17
And What About Power ?
18
Future Research

• What comes next ? To look into issues like
• Behavior of the uplink/downlink load from the
  channel efficiency point of view
• Changes in the quality of connection with the
  changing mobile users distance from the Base
  station (near-far fairness).
• Behavior of the rate during the whole life of a
  connection.
• Behavior of the signal quality (BER) during the
  whole life of a connection.
• Call Dropping and Blocking Probabilities.
• Possible Applications of Adaptive Learning
  Techniques.

19

• THANKS !