Slide 1 E3E3 ICC Beijing 21 May 2008 Simulated Annealing-Based Advanced Spectrum Management Methodology for WCDMA Systems Jad Nasreddine Jordi Pérez-Romero Oriol Sallent Ramon Agustí

Slide 2 E3E3 ICC Beijing 21 May 2008 Outlines Introduction Proposed approach Simulation and results Conclusions

Slide 3 E3E3 ICC Beijing 21 May 2008 Actual Spectrum Allocation Actual spectrum allocation is space and time invariant Traffic distribution is non- homogenous in time and space A large amount of the spectrum is underutilized Several bands are saturated A large amount of the spectrum is underutilized Several bands are saturated Decrease in the offered QoS to users  Decrease of operators profits Increase in a RAT traffic  Network infrastructure extensions and cost Decrease in the offered QoS to users  Decrease of operators profits Increase in a RAT traffic  Network infrastructure extensions and cost * FCC, Spectrum Policy Task Force “Report of the Spectrum Efficiency Working Group,” November 15, Spectrum access and not spectrum scarcity reduces spectrum efficiency*

Slide 4 E3E3 ICC Beijing 21 May 2008 New paradigm of Spectrum Management More flexible spectrum management: Advanced Spectrum Management (ASM) –Better spectrum allocation to operator –Better spectrum allocation to RATs in each operator –Better spectrum assignment to cells in each RAT A need to estimate the required amount of spectrum per cell/area

Slide 5 E3E3 ICC Beijing 21 May 2008 Problem Formulation PROBLEM  Identify the best (medium to long-term) spectrum assignment to cells –Best spectrum assignment: Maximize the spectrum efficiency while the required QoS levels are guaranteed Release some blocks of spectrum when possible –Creating Spectrum holes SOLUTION  Use a smart radio indicator that is able to reflect both macroscopic and microscopic properties of the radio network –smart radio indicator: coupling matrix

Slide 6 E3E3 ICC Beijing 21 May 2008 Objectives Considered system –Several carriers are available –Uplink of WCDMA Objectives –Estimate the number of carriers needed by a WCDMA system –Estimate the number of carriers needed by each cell –Smartly distribute the available carriers among cells in order to increase system performance –Spare carriers that could be exchanged between different RATs or operators without a risk of high interference

Slide 7 E3E3 ICC Beijing 21 May 2008 Illustrative Example The same capacity Some carriers could be released for other RATs/operators or secondary market Use ASM methodology With frequency reuse 1 All carriers are used

Slide 8 E3E3 ICC Beijing 21 May 2008 Coupling Matrix Why Coupling Matrix? –Actual estimation of the required amount of spectrum per cell based on cell load estimation –does not take into account inter-cell interference  under estimation of needed carriers based on inter-cell interference estimation –using fixed ratio a between inter-cell and intra-cell interference (a is highly related to the carrier-to-cell allocation) –Coupling Matrix Reflects interference patterns with more accuracy Extension of Compatibility Matrix Medium and long term scale

Slide 9 E3E3 ICC Beijing 21 May 2008 The Jacobian of interference system of equations Low complexity without adding signaling traffic (use available information) Medium and long term scale (fast variations such as fading are averaged) Coupling Matrix Depends on path losses, spreading factors and E b /N 0 s Coupling Matrix Total received power S l,j : indicator of the influence of mobiles in cell l over cell j that depends only on mobile path losses and services.

Slide 10 E3E3 ICC Beijing 21 May 2008 ASM Algorithm

Slide 11 E3E3 ICC Beijing 21 May 2008 Outage Probability Estimation Estimates outage probability without testing them on real systems –Otherwise, unacceptable performance for significant periods of time –Based on matrix inverse –Require path loss distribution –complexity : O(K 3 )

Slide 12 E3E3 ICC Beijing 21 May 2008 Simulation Model Outage Probability threshold 0.05 System with 3 carriers Uplink Heterogeneous traffic distribution R b = 12.2 Kbps Compared algorithms: –Uniform algorithm –ASM

Slide 13 E3E3 ICC Beijing 21 May 2008 Release several carriers Results (1/2) Outage probability constraint is satisfied: always less than 0.05 Approximately multiply by two the spectrum efficiency keeps the variation of cell’s spectrum efficiency as low as the discreet values of the bandwidth allows

Slide 14 E3E3 ICC Beijing 21 May 2008 Conclusions Introduction of a new ASM methodology: –Objectives: An efficient spectrum utilisation of licensed spectrum bands –In accordance with the existing load levels Releasing some carriers for a secondary usage in large geographical areas –When the load levels are low enough –Based on a simulated annealing –Uses coupling matrix The results show that –The QoS levels (outage probability) are respected –The spectrum efficiency is significantly increased –Some carriers are released Future work –Utilization of the released carriers by cognitive radios without polluting WCDMA users with harmful interference

Slide 15 E3E3 ICC Beijing 21 May 2008 Thank you Acknowledgment: This work was performed in project E 2 RII/E3 which has received research funding from the Community's Sixth/Seventh Framework programme. This paper reflects only the authors' views and the Community is not liable for any use that may be made of the information contained therein. The contributions of colleagues from E 2 RII/E3 consortium are hereby acknowledged.