Performance evaluation of multicast video distribution using lte a in vehicular environments (vtc fall)

Performance Evaluation of Multicast
Video Distribution using LTE-A in
Vehicular Environments
Jayashree Thota
(jaya.thota@bristol.ac.uk)
University of Bristol, UK
VTC Fall 2016, Canada
18th Sep 2016

Presentation Outline
 Motivation (Problem, Solution, Application)
 Cross-Layer Simulator - Video broadcast at high speeds
 Results Analysis
 Case study – Realistic channel Bristol, UK
 Conclusions
2

Motivation
 Vehicular Applications such as on-board internet, media downloading, map
update, video communication require
 high bandwidth
 high reliability
 low latency
 Long Term Evolution (LTE-Advanced) for Vehicular applications provides
 Faster deployment
 Reduced costs
 Higher data rates
3

Motivation
 However, LTE-A for Vehicular applications can be challenging due to high
Doppler spread and low delay requirements.
 One way to provide end to end reliability is to use systematic Raptor codes for
broadcast as any symbol can be used to decode the missing symbols.
……………. …………….
K (Source Symbols) R (Repair Symbols)
 Raptor Code provides flexibility as repair symbol generation is reconfigurable
on-the-fly depending upon channel conditions .
 Reliability can come at the expense of bandwidth.
1 2 K K+1 K+2 N=K+R
Fig 1: Raptor Codes
4

Motivation
 A cross-layer system based on latest RaptorQ codes to transmit high data rate
video with a user moving at 50kmph in a LTE-A realistic outdoor environment
 A range of MIMO modes such as
 STBC (Space Time Block Coding) – high reliability, low data rates.
 SM(Spatial Multiplexing) –high data rates depends on channel conditions.
 STBC vs SM + systematic raptor codes in application layer for various
Modulation and Coding schemes (MCS) and spatial channel correlations .
 A link adaptation model with optimised cross-layer parameters is proposed
under different channel conditions and quality of service requirements.
5

Cross-Layer Simulator
 Developed cross layer simulator comprising
1. Video simulator
• Models transmission of an H.264 video sequence
• Video encoder translates video frames in to fixed size Network
Abstraction Layer Units (NALs).
• 1 NAL = 1 UDP/RTP
2. Raptor system
3. LTE–A simulator
4. Channel simulator
6

 Developed cross layer simulator comprises
 Video simulator
 Raptor Q (IETF RFC 6330) [1]
• Raptor encoder constructs source blocks comprising ‘K’ source
symbols, each size ‘T’ bytes.
• ‘K’ source symbols followed by ‘R’ repair symbols sent through
channel for various MCS, MIMO modes and SNRs.
• Raptor decoder collects all ‘N’ symbols for each source block and if
N>K , it is successfully decoded for that source block.
 LTE-A simulator
 Channel simulator
[1] M. Luby, A. Shokrollahi, M. Watson T. Stockhammer, L. Minder, “RaptorQ Forward Error Correction
Scheme for Object Delivery,” IETF RFC 6330, Aug. 2011.
7

 Video simulator
 Raptor system
 LTE-A simulator
 IP+PDCP+RLC+MAC headers
 No ARQs in MAC layer
 No segmentation in RLC layer.
H.264 Video NALs
RTP/UDP packets
Raptor FEC broadcast streaming
framework
IP Multicast/broadcast
PDCP (header compression)
RLC (No segmentation)
MAC (No ARQs)
Physical Layer (Spatial Multiplexing
MIMO)
Table 1: Cross-Layer Simulator
8

LTE-A Physical layer
• System level simulation performed by detailed LTEA Physical layer simulator
depending upon the transport and physical channel processing.
Parameter Values
Bandwidth 10MHz
BS Transmission Power 43dBm
MIMO antenna spacing BS (10λ) , UE (0.5λ)
Antenna type Measured Pattern
Carrier Frequencies 2.6 GHz
Wireless Channel Model Extended 3D 3GPP/ITU
channel model
Mobility speeds 50 km/h
Channel Estimation 2D MMSE
Channel Sampling
frequency
15.36 M samples/s
Packet Size 1500 Bytes
LOS condition NLOS
• A 2D MMSE(Minimum Mean
Square Error) channel
estimator is used to estimate
the channel frequency
response from the pilot
structure of the LTE resource
grid.
Table 2: LTE-A system Parameters
9

 Video simulator
 Raptor system
 LTE PHY bit level simulator
• 3GPP/ITU channel model generate packet error traces
• Measured antenna radiation patterns
10

3GPP/ITU 3D channel model
• The 2D 3GPP/ITU channel model is a geometry based stochastic channel
model. It specifies the direction of rays and randomly draws the large scale and
small scale parameters. Statistical distributions are extracted from the channel
measurements.
• In this study, the existing two-dimensional (2D) 3GPP/ITU channel model,
which only focuses on the propagation in the azimuth plane, is enhanced by
extending the channel Large Scale Parameters using a 3-dimensional (3D) ray
tracer engine.
• The 3D ray tracer statistics can be imported directly into the 3GPP/ITU process
for generating 3D channel realizations, which consider propagation in both
azimuth and elevation plane. This implemented 3D ITU channel model has
been validated against a 3D ray tracer in [2].
[2] R. Almesaeed, A. Ameen, A. Doufexi, N. Dahnoun and A. Nix, “A Comparison Study of 2D and 3D ITU Channel Model,”
IEEE Wireless Days, 13-15 Nov 2013, Valencia, Spain.
11

3GPP/ITU 3D channel model
Scenario
Selection
Urban macro
Urban micro
Indoor
Outdoor
Out2in
Etc..
Large Scale
Parameters
-DS, AS, K
- Shadowing
- Path Loss
Antennas
- # of elements
- orientations
Field patterns
Multi-path
parameters
-power
- delay, AOA
- AOD etc
Network Layout
-BS and MS
Locations
- Velocities
Channel
coefficient
generation
User defined parameters
Propogation parameter generation ChIR generation
ChIR
Fig 2: Channel Generation in a 3GPP/ITU Channel Model
Extending the channel LSPs using a 3-dimensional (3D)
ray tracer engine for city of Bristol, UK
12

Antenna Patterns
• Measured Macro BS and UE antenna pattern are integrated with 3D ITU. All
patterns are 3D and include full phase and polarization information.
(a) Macro BS antenna (b) UE handset antenna
Fig 3: Total Power Measured Radiation Pattern
13

Performance Evaluation
 Cross-Layer simulator is used to broadcast a constant bit-rate video sequence
for a user moving at 50kmph using SM MIMO in urban location for
 Different SNR values (range 0dB-40dB)
 Channel Correlations (Low, High, Medium) can be characterized by
𝐻𝑑𝑒𝑡 = det ℎ𝑖𝑗 ℎ𝑖𝑗
𝐻
, where ∙ 𝐻
denotes Hermitian function, ℎ𝑖𝑗
represents channel matrix for i=1,2 for receiver and j=1,2 for transmitter
 Various Modulation and Coding Schemes
 Different Code Rates 0.95 ≤ 𝐶𝑅 ≤ 0.5 , where
• CR=
𝐾
𝐾+𝑅
; CR = 1 No Raptor ; CR = 0.5 (50% overhead)
 A link adaptation system is proposed whose inputs are SNR and Hdet and
outputs the optimum MCS and CR for a given Quality of Service (Packet Error
Rate <1% assumed in this study)
14

Results - High Correlation Channel
 It can be seen that depending upon MCS and CR there can be as much as 4dB
SNR improvement after using raptor codes to achieve a PER<1% .
Fig 4: UDP Packet Error Rate SM and STBC before raptor Fig 5: UDP Packet Error Rate SM after raptor
15

Case Study
Fig 6: Route of Bristol city Centre for case study
Fig 7: Received SNR through the route
Fig 8: Received Hdet through the route
BS
16

Case Study
Fig 9: Achievable Peak Rate for STBC without raptor and SM with raptor to achieve PER<1%
 Timeslot = 85-280sec, SM+ raptor outperforms STBC, whereas timeslot=20-85sec
STBC is better than SM. This is because STBC performance depends upon SNR,
however SM performance depends upon SNR and channel correlation.
17
SM Raptor 𝑃𝑒𝑎𝑘 𝑅𝑎𝑡𝑒 𝑀𝑏𝑝𝑠 = 1 − 𝑃𝐸𝑅 ∗ 𝑏𝑖𝑡𝑟𝑎𝑡𝑒 ∗ 𝐶𝑅

Conclusions
 A link adaptation system is proposed whose inputs are SNR and Hdet(channel
conditions) and outputs the optimum MCS and CR for a given Quality of Service.
 Spatial Multiplexing (SM) MIMO + application layer systematic raptor codes
 can provide up to 4dB SNR improvement to achieve PER=<0.01 in highly
correlated channels (bad-channel conditions) . This requires a high number
of repair symbols to provide reliability and reduces transmission efficiency.
 outperforms STBC and can be used to increase the reliability and
transmission efficiency of the system in lower correlation (good channel
conditions) .
18

I would like to thank my supervisors for their guidance, EPSRC and Jaguar Land
Rover for sponsoring this study.
Thank You
{ jaya.thota, berna.bulut, a.doufexi, s.armour, andy.nix}@bristol.ac.uk
Any Questions?
19

Performance evaluation of multicast video distribution using lte a in vehicular environments (vtc fall)

More Related Content

What's hot (19)

Similar to Performance evaluation of multicast video distribution using lte a in vehicular environments (vtc fall) (20)

More from Communication Systems & Networks (17)

Recently uploaded (20)

Performance evaluation of multicast video distribution using lte a in vehicular environments (vtc fall)