PAGE 1 PAGE 1

Heterogeneous networks for dependable wireless access

and the 1000x capacity challenge Valentin Gheorghiu

PAGE 2

• Why 1000x?

• How to achieve 1000x?

• Heterogeneous networks and challenges

LTE-Adv Rel.10/11 eICIC/FeICIC (ICIC=Inter-cell interference coordination)

Mobility challenges and dependability

Hyper dense HetNets

– Network management with UltraSON

Future mobility challenges

RF challenges

• Concluding remarks

Summary

PAGE 3

Strong Mobile Data Demand

0

6

12

2011 2012 2013 2014 2015 2016

Ex

ab

yte

s p

er

Mo

nth

Overall Mobile Data Traffic Growth

NETWORK CAPACITY CHALLENGE!

0.6 EB 1.3 EB

4.2 EB

2.4 EB

10.8 EB

6.9 EB

being planned

by operators

over next decade

PAGE 4

Capacity increases enable by topology changes

Evolved 3G (EV-DO Rev. B & HSPA+)

Data optimized 3G (EV-DO & HSPA)

3G (IMT-2000): Voice & Data (e.g. CDMA2000 1X & WCDMA)

2G: Voice Capacity (Digital e.g. GSM & IS-95)

1G: Voice (Analog e.g. AMPS)

Next Generation

Leap

Next Generation

Leap

Next Generation

Leap

Radio link

approaching

theoretical limit

LTE (OFDMA)

Different dimension of improvement:

adding small cells like picos and

femtos and mitigate interference

Bring network closer to users and

leverage Hetnets1—add small

cells—for next leap in performance

1Leveraging heterogeneous network topology: macro network with added small cells like picocells and femtocells (cells

with different transmit power)

PAGE 5

We Can Reach The Air Link Limit—Shannon’s Law

Still ways to improve system capacity

Capacity ≈ n W log2(1+ ) Signal

Noise

Number of

Antennas

More

Spectrum E.g. Mitigate

interference

Qualcomm proprietary and confidential

PAGE 6

The Biggest Gain—Re-Use Shannon’s Law

Capacity ≈ n W log2(1+ ) Signal

Noise

Number of

Antennas

More

Spectrum E.g. Mitigate

interference

Qualcomm proprietary and confidential

PAGE 7

• Conventional cell splitting is an inefficient way to address non-uniform and exploding data demand

In addition, site acquisition is difficult and cells already approaching lower limits of site to site distance

Heterogeneous Networks - Overview

Heterogeneous deployments with eICIC1 and feICIC1

(LTE Rel.10/11)

Hyper Dense Heterogeneous deployments with Small Cells (LTE Rel.12~)

Homogeneous deployments (Macro

only, e.g. LTE Rel.8)

Co-channel deployment

Cell Range Expansion

Adaptive Resource

Partitioning

Almost blank subframes

Advanced receivers with

interference cancelation

More spectrum (higher frequency

bands, e.g. 3.5GHz~)

Dual connectivity

Self Organizing Networks (UltraSON)

Femto Pico

Pico

Pico Macro

1 ICIC=Inter Cell Interference Coordination, e=Enhanced, Fe=Further Enhanced

PAGE 8

• Resources between macro BS and low power BSs split in the time domain

• Range Expansion increases footprint of co-channel Picos/Femtos

Allows more UEs to be served by low power BSs

More equitable distribution of capacity among Macros and Picos/Femtos

• Full Range Expansion can be supported in LTE-Advanced

Large bias to compensate the power difference between Macro and Pico

Enabled by resource partitioning and enhanced UE receiver

Heterogeneous Networks and Cell Range Expansion

Macro

Limited footprint of Picos due to Macro signal

Pico Macro Pico Pico Pico

Increased footprint of Picos when Macro

frees up resource

In subframes reserved for Picos In subframes reserved for Macros

Example:

Semi-static allocation 50%

Macro and

50% Picos

PAGE 9

• Mobility performance is extremely important from a network dependability point of view

• Number of handovers is expected to increase because of the increase in the number of cells

Mobility management becomes a problem

• Simulation results show big increase in the number of handover failures, specially in certain scenarios (e.g. Pico-Macro)

System simulation based on scenarios used in 3GPP

Mobility Challenges and Dependability (1)

Simulation assumptions

• 30km/h

• Macros deployed at 500m distance

• 30picos/macro cell

• Fully loaded system

• HO Failure based on service outage

(device goes into link failure) 0

10

20

30

40

50

60

Baseline (No eICIC) Rel.11 9dB CRE w/ resourcepartitioning

Handover Failures (%)

Macro-Macro

Macro-Pico

Pico-Macro

Pico-Pico

PAGE 10

• CRE and resource partitioning help mitigate some of the mobility problems

• Pico-Macro HO is more likely to fail than Macro-Pico HO

The Pico signal strength varies fast because of shadowing and much lower antenna height

Sudden change in the Pico’s signal due to mobility would introduce similar challenge for both inbound mobility and outbound mobility

However, only a small fraction of Macro-Pico handovers go through the outage area

On the other hand, all outbound handovers are subject to link failure because of strong interference from the macro

Mobility Challenges and Dependability (2)

• Mitigation Techniques

Different HO parameters for different cell types

Differentiate devices based on speed ⇒ high speed devices kept only on macro cell

• Devices have little freedom as mobility is driven by the network

PAGE 11

• More capacity improvements than what is offered by LTE Rel.10/11 will be needed to fulfill 1000x

• Hyper densification is likely to be the key to reach 1000x capacity increase

Small cells deployed in higher frequency bands

Hyper Dense HetNets

METRO

RESIDENTIAL

ENTERPRISE

USER DEPLOYED

Typically indoor small cells

OPERATOR DEPLOYED

Indoor/outdoor small cells1

Extreme Densification—Small Cells Everywhere

4G Relays

& Wireless

Backhaul

1 Such as relay and Pico/Metro/RRH small cells for hotspots. RRH= Remote Radio Heads, in addition Distributed Antenna Systems are used in HetNets

PAGE 12

• Network management has to be automated such that deployment is scalable⇒ Self Organizing Networks(SON) techniques will be very important

• Basic SON features standardized from Rel.8 (e.g. ANR)

Hyper Dense Hetnets – Network management

Self Configuration: Automatic cell

parameter and backhaul config.

• Automatic Cell ID (PCI) selection

• Automatic neighbor discovery (including

3GPP ANR)

Backhaul Aware Operation:

Handle backhaul constraints • Backhaul quality aware load balancing

Category UltraSON Features

Mobility Management: Optimize

HO performance and reduce

signaling load

• Frequent Handover Mitigation

• Forward handover

• Robust mobility (including 3GPP MRO)

Dynamic Resource and Tx Power

Management: Optimize capacity,

minimize pilot pollution and load

balancing

• Tx power management

• Resource partitioning and coordination

(including 3GPP ICIC)

• Load balancing

PAGE 13

• Mobility management will be more challenging because of the increase in number of cells and frequencies

• Dual connectivity will be one of the solutions

Devices are connected to a Macro cell and a small cell at the same time

The control information(including mobility control) will be transmitted on the Macro layer

Connectivity is maintained on the macro cell while the small cells are used opportunistically for traffic offload

Increase overall dependability

Hyper Dense Hetnets – Mobility and Dependability

• Mobility state detection

Network can make “smart” decisions on whether to handover a device to a small cell based on its mobility

PAGE 14

• Higher spectrum bands suitable for small cells

• To achieve the needed increase in capacity more spectrum will be needed

RF circuits will have to support a very wide range of frequency bands

– 450MHz~60GHz

Broadband processing for carrier aggregation

– ~100MHz bandwidth

RF Challenges – Dependable RF

~450 MHz

60GHz

~3GHz

Very High Bands

enable Access In

Every Room

Wide Area

Spectrum

INDOOR

HOTSPOT

3.4 to 3.8 GHz Emerging as a new

small cell band

PAGE 15

Is 1000x really possible?

A SCALABLE APPROACH TO MEET THE 1000X CHALLENGE

Downlink Median Throughput Gain* Hyper-dense cells and more spectrum over Macro-only deployment

0

250

500

750

1000

1250

1500

0 72 144 216

1000x Capacity Gain Using 144 Small Cells Per

Macro and 10x Spectrum**

Number of Small Cells Per Macro

x

x

x

x

x

x

36

* This example shows one possible combination of small cells and spectrum to achieve 1000x capacity gain. ** Macrocells use 10MHz spectrum in 2GHz and small cells use 100MHz spectrum in 3.5GHz, 200 active users per macro cell.

Re

lati

ve C

apac

ity

Gai

n

PAGE 16

• Mobile Data Traffic is expected to grow 1000x over the next 10 years

• Dependable networks that deliver the data timely and maintain seamless connectivity are required

• Heterogeneous network deployments will be the key to meeting the data traffic demand

Rel. 10/11 Hetnets enabled through resource partitioning and advanced receivers

Hyper Dense Hetnets with small cells on higher frequencies, dual connectivity

• Dependable RF circuits able to process frequencies up to 60GHz will be needed

Spectrum from 450MHz up to 60GHz will be used, small cells are very likely to be deployed in the higher frequency bands

RF circuits capable of processing up to 100MHz bandwidth will be required for carrier aggregation

Concluding Remarks

PAGE 17

Thank You

Will There Be 1000x Demand? It’s Just a Matter of Time…