Coordinated Multipoint (CoMP) Transmission for LTE-Advanced Networks in Dense Informal Settlements

Coordinated Multipoint (CoMP)Transmission for LTE-Advanced Networksin Dense Informal Settlements

Beneyam B. Haile, Edward Multafungwa and Jyri HämäläinenDepartment of Communications and NetworkingSchool of Electrical Engineering, Aalto UniversityEspoo Finland

IEEE AFRICON 2013, Mauritius, September 11

OutlineØ MotivationØ Deployment scenarioØ CoMP criteria and schemeØ System simulation parameters and assumptionsØ Simulation resultsØ Conclusion

2

Motivation (1/2)Future mobile networks are expected to deliver consistentthroughput experience throughout the network

3

This is challenging in LTEnetworks due to varying desiredsignal strength, intercellinterference and cross-tierinterference across the network.

CoMP technology is considered as a candidate solution andit has recently been studied by academia and industry.

More challenging in denseinformal settlement area

Motivation (2/2)CoMP performance is evaluated in 3GPP release 11 assuming idealbackhaul

4

An ongoing work item on evaluation of CoMP performanceassuming practical backhaul is there in 3GPP release 12

Simulation works assume uniform cellular layout

This study evaluates performance ofa practical CoMP technique in anexamplary dense informal settlementarea considering terrain and 3Dbuilding information .

Deployment scenario in an example informalsettlement areaThree one-sectored macro celldeployed in Hanna Nasif area,Dar es Salaam, Tanzania

5

Hanna Nasif: around 40,000people/km^2, 3000 buldings

Users (LTE UEs) are droppedrandomly in the simulationarea.

Note: Instead of users, we canalso have hotspot nodes usingCoMP as a backhaul

CoMP criteria and scheme 1(2)Coordination is assumed among the three macro cells

6

CoMP criteria and scheme 2(2)

7

CoMP UEs are selected based on the average of the differencebetween received signal powers from cell 1 and cell 2, and cell 1and cell 3:

CoMP technique: Quantized Co-phasing (QCP), suboptimal andpractical (used in WCDMA/HSPA/LTE)

Where

System simulation parameters andassumptions

Parameter Values/AssumptionsAir Interface LTE FDDCarrier Frequency 2000 MHzSimulation Radio propagation modeling (WinProp)

Static system level simultion (Matlab)5 m resolution; 1.5 m prediction height

CoMP Three cooperating cellsDownlink JP CoMPQ-CP limited feedback with N = 3 feedbackbits

Macro Cells Cell 1 Cell 2 Cell 3Transmit Power 46 dBm 46 dBm 46 dBmAntenna Height 15 m 20 m 25 mAntenna Pattern Kathrein

741984Kathrein741984

Kathrein741984

Azimuth 140 0 250Intersite distance Approximately 500 mMean UE Number 20UE mobility Mobility off; Users distributed randomly in

the simulation area in each snapshotFading Shadow fading: Gaussian (0 dB mean, 8 dB

standard deviation), decorrelation distance50mFast fading: Zero mean complex Gaussian

Buildings Buildings of variable dimensionsBuilding heights 3-6 mBuilding benetration loss: 10 dB

8

3GPP guidlines are used

Average received power is computedusing dominant path model implementedin the WinProp propagation modelingtool

SINR performance result (1/2)

9

CoMP threshold (dB)1 3 6 10

0.5% 2.5% 15% 36%

The higher the CoMP threshold thelarger the number of CoMP UEshence, the better the overall SINRperformance

Percentage of CoMP UEs

One third of power transmission

At low CoMP thresholds (≤ 3dB), theoverall SINR performance improvementrelative to non-CoMP case is negligible

SINR performance result (2/2)

10

CoMP threshold (dB)1 3 6 10

0.5% 2.5% 15% 36%

The difference in transmission powershas very little impact on overall SINRperformance

Percentage of CoMP UEs

Full power transmission

Most of the SINR gain is achievedfrom avoiding the interferencesthan the Q-CP transmission technique

SNR gain performance result

11

CoMP threshold (dB)1 3 6 10

0.5% 2.5% 15% 36%

The best SNR gain performance isachieved for the smallest threshold,thus smallest power imbalance

Percentage of CoMP UEs

Full power transmission

Similar trends in SNR gains are notedthrough analytical studies

Spectral efficiency performance result (1/2)

12

Full power transmissionOne third of power transmission

The larger the CoMP threshold, the higher the overall spectral efficiencygains

Spectral efficiency gain performance result (2/2)

13

Full power transmissionOne third of power transmission

Energy efficiency gains are possible for one third power transmissionwithout considerable reductions in spectral efficiency gains

The smaller the CoMP threshold, the higher the CoMP UE spectralefficiency gains

ConclusionsCoMP enhancements is a viable option for throughputenhancements in LTE networks that are being deployed in denseurban areas

14

The higher the threshold, the larger number of CoMP enabledUEs but smaller spectral efficiency gains for CoMP UEs

Tradeoff between energy efficiency and spectral efficiency canbe managed through optimal selection of transmission power ofcooperating nodes.

Future research: further study of energy-efficiency trade-off,CoMP in HetNets in densely populated area, CoMP as a backhaulof small cells and relays.

15

Thank you for your kind attention!

Further info:Beneyam Haile (Doctoral Researcher) and Edward Mutafungua (PostDoctoral Researcher)Aalto UniversitySchool of Electrical EngineeringDepartment of Communications and NetworkingOtakaari 5A, Espoo, [email protected] and [email protected]: +358 44 2108323 and +358 40 733 3397

16

Support Slides

Why Focus on Informal Settlements?• Focus of on informal settlements (slums) in suburban/urban areas

– Emerging markets have fastest slum growth rates; 30-50% of globalurban population in 2030 (UN HABITAT 2007)

– Characterized by very high population density (>4000 people/sq km)and low income (1-3 USD/day)

– Underserved: Limited access to key services (electricity, sanitation,healthcare, broadband etc.)

Kibera, Nairobi, Kenya(pop. 230000 – 1 million, area 2.5 sq km)

Korail, Dhaka, Bangladesh(pop. 120000, area 0.4 sq km)

Dominant path model• Dominant Path Model

– Faster computation time than ray tracing models– Models dominant path between TX and RX pixel– More accurate than COST 231 model in scenarios with strong multipath

propagation– Combination of urban and indoor predictions possible (CNP mode).

• Urban Dominant Path Model (UDP) for outdoor• Indoor Dominant Path Model (IDP) with a higher resolution for indoor• Potentially good choice for urban or densely built suburban scenarios, particularly for

cases with below rooftop transmitters (small cells)

Comparison of different approaches(COST 231, Ray Tracing, DPM)

Average received power map

Qiantized co-phasing