Range Assignment for Power Optimization in Load-Coupled Heterogeneous Networks

Lei You, Lei Lei, and Di Yuan (speaker)

Linköping University

Sweden

IEEE International Conference on Communication Systems (ICCS)

Macau, Nov. 19-21, 2014

Outline

l  Introduction

l  Theoretical results

l  Proposed algorithm

l  Performance evaluation

l  Conclusions

2

Heterogeneous Networks (HetNets)

3

l  Heterogeneous  Network:    overlaying  macro  cells  (MCs)    +  underlying  small  cells(SCs),  also  refer  to  low-­‐power  nodes  (LPNs)  

l  LPNS  offload  data  traffic  from  macro  cells  and  provide  enhanced  capacity  to  traffic  hotspots  

l  If  each  user  equipment  (UE)  selects  its  serving  cell  by  the  best  received  signal  power  -­‐-­‐  some  cells  may  be  over  loaded  

?

l  How  to  opImize  UE-­‐cell  associaIon?    Via  cell-­‐specific  offsets  on  LPNs  

Cell’s Offset and Load-Coupled HetNets

4

●  Cell’s  offset:  a  virtual  amplificaIon  in  UE-­‐cell  selecIon,  does  not  physically  change  transmit  power.  

●  Each  UE  follows  the  rule  of  selecIng  the  cell  with  the  best  received  signal  power  plus  an  offset  

 

●  Cell’s  load:  the  fracIonal  usage  of  the  Ime-­‐frequency  resource  units  

●  Load  coupling:    refers  to  the  dependency  relaIon  between  the  cell  load  levels  due  to  mutual  interference.  

Load in cell k Load in cell i

Energy Minimization Problem

5

●   What  is  the  op7mal  opera7ng  

             load  for  each  cell?  

Energy Minimization

Cell Load Coupling

LPN Offset Assignment

●   What  is  the  op7mal  offset  level  

             for  each  LPN?  

●  Energy:  the  product  of  cell’s  load  and  power  

●  Cell’s  load  level  affects  the  sum  transmission  energy    

●  LPN  cells’  offset    à  UE-­‐cell  associaIon  à  cells’  load  levels  à  energy  consumpIon  

6

Problem Formalization

Demand

7

Theoretical Characterizations

●  Full  load  is  op7mal  in  energy  minimizaIon  

             -­‐-­‐  ReducIon  from  Maximum  Independent  Set  (MIS)        -­‐-­‐  The  complexity  jusIfies  the  development  of  subopImal                soluIon  

●  The  energy  minimizaIon  problem  is  NP-­‐hard  

Tabu Search for Offset Optimization

8

Initialize a feasible offset solution

UE-Cell association (+ offset)

Optimal power vector, at full load

Stop rule is triggered ?

Yes: Algorithm stops, return the offset solution

No: Update offset and Tabu list

●  Propose  Tabu  Search  for  compuIng  offset  to  minimize  energy  at  full  load.  

Simulation Setup

9

−3 −2 −1 0 1 2 3−3

−2

−1

0

1

2

3

1

2 3

4

5 6

7

89

1011 12

13

1415

16

17

1819

20

21l  7  hexagonal  cells,  where  the  center  is  

deployed  with  an  MC  

l  Randomly  place  two  LPN  cells  in  each  MC  

l  30  UEs  for  each  cell,  randomly  and  uniformly  distributed  

l  The  simulaIon  parameters  follow  the  LTE  standard  

            LPN: 50 mW

Numerical Results

10

In  average,  the  energy  consumpIon  for  op3mized  offset  is  about  31.9%  lower  than  that  zero  offset.

Numerical Results (cont’d)

11

By  using  the  opImized  offset  to  LPNs  to  adjust  the  cell-­‐UE  associaIon,  the  UEs  are  assigned  to  more  suitable  cells.    

Energy consumption in both MC 3 and 6 are significantly reduced in the proposed solution (optimized offset)

Conclusions

12

l  An  energy  opImizaIon  framework  for  load-­‐coupled  HetNet  via  LPN  range  assignment.

l  TheoreIcal  results:  l  With  cell  power  constraint,  operaIng  at  full  load  is  opImal  

l  The  energy  opImizaIon  problem  is  NP-­‐hard  

l  Propose  a  Tabu-­‐search  based  algorithm  to  opImize  cells’  offsets  

l  The  total  energy  consumpIon  is  significantly  reduced  by  using  the  opImized  offset  soluIon.  

www.liu.se