Alleviating cellular network congestion caused by traffic lights Hind ZAARAOUI, Zwi ALTMAN, Tania...
-
Upload
john-walsh -
Category
Documents
-
view
217 -
download
0
Transcript of Alleviating cellular network congestion caused by traffic lights Hind ZAARAOUI, Zwi ALTMAN, Tania...
Alleviating cellular network congestion caused by traffic lights
Hind ZAARAOUI , Zwi ALTMAN, Tania JIMENEZ, Eitan ALTMAN
interne Orange2
II.
III.
IV.
V.
Definition of the problem
Mobility model
Simulation and results
Resource allocation small-macro cells
Radio modelling
I.
VI. Conclusion and ongoing work
interne Orange3
I. Definition of the problem
𝝀𝟏
Macro cell
Small cell
𝝀𝟐
interne Orange4
II. Mobility model
Overtaking is excluded in the simulation:
The algorithm used in the simulation is therefore:
interne Orange5
III. Radio modelling: instantaneous cell load definition
Throughput of a user located at using modified Shannon formula:
The load density in stationary fixed case ,
The load density in non stationary mobility case
The instantaneous cell load for discrete case
interne Orange6
III. Radio modelling: network performance indicators
Instantaneous mean user throughput:
File transfer time for a user :
interne Orange7
IV. Resource allocation small-macro cells (1/2)
Macro-cell only
𝑚𝑎𝑐𝑟𝑜 :𝑤𝑠𝑚𝑎𝑙𝑙 :∅
Full frequency reuse (macro &
small cells)
𝑤𝑤
Dynamic frequency bandwidth
splitting (macro & small)
(1−𝛿 (𝑡 ) )𝑤
𝛿(𝑡)𝑤
Mean optimal frequency
bandwidth split (macro & small)
(1−𝛿∗ )𝑤
𝛿∗𝑤
interne Orange8
IV. Resource allocation small-macro cells (2/2)
Dynamic frequency bandwidth splitting :
For the case of (proportional fair), we maximize this utility by finding the solution of :
Mean optimal frequency bandwidth split:
interne Orange9
Network and traffic characteristics
interne Orange10
V. Simulation and results (1/2)
Loads comparison
MO
MFqSFqS
Reuse
interne Orange11
V. Simulation and results (1/2)
Mean user throughput in time and mean file transfer time comparison
FqS solution provides a significant gain with respect to MFqS: taking MFqS as reference
interne Orange12
Conclusion and ongoing work
interne Orange13
Scheduling in presence of mobility (1/2)
Normal scheduling
t = 1
scheduling percentile relative to the mobile user = 50%
Normal scheduling
t = 2
scheduling percentile relative to the mobile user < 50%
Dynamical scheduling
t = 1
scheduling percentile relative to the mobile user > 50%
Dynamical scheduling
t = 2
scheduling percentile relative to the mobile user = ?
interne Orange14
Scheduling in presence of mobility (2/2)
1. Normal Scheduling:
- Takes into account only past and present;- The utility function to maximize (-fair scheduler):
: mean past data rate: scheduler factor (variable that maximizes the utility): present data rate that can be received
2. Dynamical Scheduling:- Takes into account future and present;- The utility function to maximize (-fair forecast scheduler):
Multilevel beamforming in mobility scenarios
Context: Massive MIMO technology evolves rapidly towards antenna arrays with larger size, allowing to support multilevel beamforming
Multilevel beamforming is based on hierarchical beam structure which reduces the scheduling complexity
Objective: adapt multilevel beamforming to different mobility scenarios
z
x
yq
- f
f
dx
dz
Thank you for your attention