EFFECT OF TRANSMIT AND RECEIVE ANTENNA DIVERSITY ON A HIGH DATA RATE PACKET CDMA WIRELESS SYSTEM

1
EFFECT OF TRANSMIT AND RECEIVE ANTENNA DIVERSITY
ON A HIGH DATA RATE PACKET CDMA WIRELESS SYSTEM

• Anna K Dinnis and John S Thompson
• School of Engineering and Electronics
• The University of Edinburgh
• trina.dinnis_at_ee.ed.ac.uk

Sponsored by EPSRC and Nortel Networks
2
Contents

• Introduction
• The 1xEV-DO system
• System model
• Results
• Conclusions

3
Electronics at Edinburgh

• 25 academics, 120 research staff/PhD students
• Top ratings for teaching and research
• 5 rating in 2001 RAE
• Main research areas
• Micro Nano systems
• IC Design/Systems on chip
• Energy systems
• Digital Communications
• Now part of the School of Engineering and
  Electronics

4
Institute for DigitalCommunications

• Centre for Communication Interface Research
  (CCIR)
• Signals Systems Group (SASG)

Further details at http//www.see.ed.ac.uk/resear
ch/IDCOM/
5
SASG Recent Activities

• Array processing
• Nortel Smart Antenna programme
• Space/time radar with BAE/QinetiQ
• Statistical signal processing
• seismic data analysis using HOS
• sonar data analysis using HOS
• DSL system analysis with BT/Fujitsu
• Biomedical signal processing
• blind ICA in EEC
• ultrasound signal processing
• Mobile VCE
• Wireless Enablers OFDM
• Personal distributed environment

• Nonlinear signal processing
• Radial basis functions
• GSM EDGE mobile radio
• radar receivers for non-Gaussian clutter
• CDMA interference rejection
• multi-user detection
• 4G system design with Elektrobit
• 3G design with LGIC
• Image processing
• wire frame modelling
• low-bit rate coding
• algorithm fusion

6
1xEV-DO

• 1xEV-DO (1xEVolution-Data Only) system developed
  for the transmission of packet data at high rates
• Based on CDMA system

7
SPEECH VS. DATA

• Speech
• constant data rate
• latencies over 100ms intolerable
• data rate traffic is symmetrical

• Packet Data
• data rate can change
• much higher latencies acceptable
• downlink rate generally higher than uplink rate

8
cdma-2000 1xEV-DO Downlink

• The available channel capacity can be exploited
  using adaptive modulation
• Channel conditions continuously monitored
• BS transmits at full power to one user at a time
• BS is more likely to transmit to a mobile when
  its channel is at its best

9
1xEV-DO

• Base stations transmit pilot signals at full
  power. Mobile stations use pilot signals to
  determine SINR from each BS

MS
BS
BS
MS
MS
10
1xEV-DO

• Base stations transmit pilot signals at full
  power. Mobile stations use pilot signals to
  determine SINR from each BS
• Mobile stations select BS with best SINR,
  determine data rate and transmit requests for
  data

MS
BS
BS
MS
MS
11
1xEV-DO

• Base stations transmit pilot signals at full
  power. Mobile stations use pilot signals to
  determine SINR from each BS
• Mobile stations select BS with best SINR,
  determine data rate and transmit requests for
  data
• Base station selects a mobile using scheduling
  and transmits to it with full power at data rate
  requested

MS
BS
BS
MS
MS
12
1xEV-DO Frame Structure
13
SINR to Data Rate Mapping
max data rate 4915.2 kbps
max data rate 3686.4kbps

• where C actual capacity of mobile

14
Proportional Fairness Scheduling
MS
MS
BS
MS
MS

• Average data transmission rate calculated for
  each MS
• Scheduler selects mobile with highest ratio of
  requested data rate to average data rate
• The more users there are, the more opportunities
  the scheduler will have to select a user when its
  SINR is well above average
• With more than one user throughput will depend
  not only on average SINR but also on how much
  SINR varies around this average

15
System Model

• MS randomly positioned in shaded area
• SINR calculated for each BS
• BS with best SINR selected
• Data rate calculated

10
6
11
9
3
2
12
4
5
1
8
7
16
Channel Model

C(n) temporal (Rayleigh) fading coefficient
for path n l distance between BS and
MS p path loss exponent ?
shadowing coefficient (standard deviation
8dB) For two path model secondary path 5dB down
17
MIMO

• Assume the transmitter knows the channel
  coefficients
• Form N orthogonal eigenvalue channels to
  receiver
• Equal power used for all channels (water
  pouring NOT used)
• Data rate found using average SINR of channels

1
1
2
2
N
M
18
SINR Cumulative Distribution Function

• flat fading
• p 4 gives substantially better SINR results
  than p 2.8

19
path loss exponent 2.8, single path

• throughput increases with number of users
• doubling number of receive antennas approx
  doubles throughput for 1 user
• increase in throughput with number of users
  becomes smaller as number of receive antennas is
  increased

20
path loss exponent 2.8, single path

• only a small improvement in throughput from
  increasing the number of transmit antennas

21
path loss exponent 4, single path

• doubling number of transmit antennas increases
  throughput by more than 50

22
SINR to Data Rate Mapping

• the larger SINR is the greater the increase that
  is required to double the throughput
• methods such as receive antenna diversity which
  increase SINR become less effective at high SINRs

23
path loss exponent 2.8, two path

• increasing the number of transmit antennas
  results in a decrease in throughput
• doubling the number of transmit antennas doubles
  the number of sources of interference

24
path loss exponent 4, two paths

• throughput for (1,2) better than for (2,2)
• (2,4) gives best performance
• throughput for (4,4) better than for (1,4)
• MIMO performs better than receive antenna
  diversity at high throughputs

25
path loss 2.8, two paths,max 256QAM (4915.2
kbps)

• throughput for (1,4) and (2,4) decreased by 10
• throughput for (4,4) unchanged, now better than
  (1,4)

26
path loss 2.8, two paths, max 64QAM (3686.4
kbps)

• throughput for (1,4) considerably reduced
• throughput for (4,4) unchanged

27
Conclusions

• Increasing number of transmit antennas does not
  always result in an improvement in throughput
• Receive antenna diversity works better for low
  SINRs
• MIMO works better for high SINRs
• factors affecting cutoff point include coding set
  and number of paths
• MIMO increases maximum possible data rate by a
  factor of N