Mobile Relay Enhancements For LTE-A University of Oulu Wireless Communications Research Seminar 2012 Simon Scott LOCON - Local connectivity and cross layer design for future broadband mobile systems
Mobile Relay Enhancements For LTE-A
University of Oulu Wireless
Communications Research Seminar 2012
Simon Scott
LOCON - Local connectivity and cross layer design for future broadband mobile systems
© Centre for Wireless Communications, University of Oulu
Introduction
• Support for fixed cellular relays was introduced in 3GPP
release 10 (LTE-A).
• Relays may be used to extend the coverage of cells, or
improve spectral efficiency by decreasing user to
infrastructure distance.
• Wireless backhaul makes relays cheaper and faster to
deploy than eNodeBs, particularly in built up areas.
2 16.2.2012
© Centre for Wireless Communications, University of Oulu
Introduction: mobile relays
• In 3GPP release 10 handover is not supported for relays.
• The 3GPP has approved a study item (36.416) on mobile
relays for release 11.
• The mobile relay study item focuses on a high speed train
scenario characterised by:
• High speed ~300Km/h
• High penetration loss through carriage walls.
• Known path
• UE are stationary, or at pedestrian speeds relative to
the relay nodes.
3 16.2.2012
© Centre for Wireless Communications, University of Oulu
Motivation for Mobile Relays
• High speed train networks are being deployed at
increasing rate worldwide.
• Providing cellular services to onboard passengers is
important but more challenging than in typical
environments:
• High penetration loss, may be upto 24dB on some
more modern trains.
• Severe doppler shift.
• Reduced handover success rate, reduced spectral
efficiency.
4 16.2.2012
© Centre for Wireless Communications, University of Oulu
Motivation for Mobile Relays
• The use of mobile relays can overcome these problems:
• Penetration loss eliminated, backhaul link antennas
on exterior of train, access link antennas on interior.
• Group mobility increases succesful handover
probability, and reduces handover signalling
overhead.
• More advanced signal processing algorithms may be
utilised in relays than in UE to combat doppler shift.
• UE may use reduced transmit power prolonging
battery life.
5 16.2.2012
© Centre for Wireless Communications, University of Oulu
Other solutions
• One solution is to deploy dedicated macro eNodeBs with
(or without) L1 repeaters onboard:
• Dedicated macro eNodeB deployment with
directional antennas provides path for all train traffic,
with larger cell overlap to increase handover success
rate.
• Such systems have been successfully deployed in for
example metro train systems.
• In addition L1 repeaters can be used to overcome the
penetration loss.
• Several proposed and implemented systems relying on
WiFi for access link.
6 16.2.2012
© Centre for Wireless Communications, University of Oulu
Advantages of mobile relays over
existing solutions
• Improved quality of service for all cellular services.
• Can use existing cellular infrastructure, reducing cost over
solutions requiring dedicated eNodeBs.
• Improved spectral efficiency over repeaters.
• Onboard services.
7 16.2.2012
© Centre for Wireless Communications, University of Oulu
Some challenges
• Backhaul link and access links may operate in shared
spectrum, duplexing operation may be required.
• Resource allocation at the DeNodeB in order to provide
sufficiently high data-rates for the the backhaul links, while
maintaining fair sharing of resources with normal macro
users.
• The backhaul link is the capacity bottleneck of the whole
system, high speed poses challenges to MIMO-OFDMA
link adaptation when considering maximizing the backhaul
link spectral efficiency.
8 16.2.2012
© Centre for Wireless Communications, University of Oulu
System Model
9 16.2.2012
ce
ll bo
rde
r
crX2 interface
moving train
L meters
RN#1
Donor eNB#X
Donor eNB#Y
X2 interface
RN#2RN#N-1RN#N
moving cooperative cell group
© Centre for Wireless Communications, University of Oulu
System level simulations
• LTE compliant system simulator modified to simulate high
speed train scenario.
• Backhaul link (DeNB -> MRN) is modelled accurately.
• Multiple backhaul links.
• Direct link from eNodeB to UE onboard train is also
modelled.
• Access link is not modelled at present. It is assumed that
the backhaul link is the capacity bottleneck, and that the
access link is always better than the backhaul link.
• Static 2:2 half-duplex operation of the MRN backhaul link.
10 16.2.2012
© Centre for Wireless Communications, University of Oulu
Simulator layout
• Trains are dropped on
a track with radius of
4Km.
• Train dropped along
the track such that the
whole train is inside the
central 57 cell layout.
• A train consists of 8
carriages 30m long,
each with an MRN.
11 16.2.2012
© Centre for Wireless Communications, University of Oulu
Simulator layout
• 20 active train users are evenly distributed amongst the
carriages of a train.
• Train users are paired with the MRN access point located
in the same carriage.
• Normal macro users are evenly distributed throughout the
cell layout.
• MRNs are paired independently with base-stations, the
serving link is determined by path-loss.
12 16.2.2012
© Centre for Wireless Communications, University of Oulu
Resource allocation
• MRNs are guaranteed a proportion of resources based on
the number of users they are serving (half duplex <= 50%).
• Proportional fair scheduling with a rate constraint is used
for normal macro users.
• Macro users are scheduled while the MRN backhaul link is
inactive (access link active).
• When users are connected through a MRN, it is assumed
that all MRNs may cooperate to share resources perfectly
fairly.
• Proportional fair scheduling is used for the users onboard
the train in the case of being connected directly to the
eNodeB.
13 16.2.2012
© Centre for Wireless Communications, University of Oulu
Simulation parameters
14 16.2.2012
System Bandwidth: 10MHz (50PRB)
Propagation Environment Urban Macro
Number of Macro Users ~10 per cell
Number of Train Users 20
Base-station Tx Antennas 1
UE Rx Antennas 2
MRN Rx Antennas 4
Carriage Penetration Loss 20dB
Traffic Model (Train users) Full Buffer
Traffic Model (Normal
Macro users) 250Kbits/s Rate Constraint
© Centre for Wireless Communications, University of Oulu
Results: Train user throughput
15 16.2.2012
© Centre for Wireless Communications, University of Oulu
Results: Normal macro user throughput
16 16.2.2012
© Centre for Wireless Communications, University of Oulu
Further work
• Extend to multiple transmit antennas in MRN backhaul.
• Access link model:
• Not a fair assumption that access link will be
bottleneck in full duplex-case.
• Allow for dynamic half-duplex scheme to be
implemented (1:3, 2:2, 3:1).
17 16.2.2012