1. MIMO-OFDM for High-Speed WLANs
Frederik Petré Bart Van Poucke
André Bourdoux Liesbet Van der Perre

2. What is multi-antenna all about?
Phased Array
MTMR
MIMO
SDMA
Smart antenna
Beamfoming
Null steering
Space-time coding
Adaptive array

3. Need for 4G High-Speed WLANs
5 m
1 Mbps
10 Mbps
100 Mbps
1 Gbps
Maximum Data rate
100 kbps
50 m
500 m
Range 1G
WLAN
802.11
1-2 Mbps 3G
WLAN
802.11a/g
6-54 Mbps 4G
WLAN
802.11n
> 100 Mbps 2G
WLAN
802.11b
Mbps
Higher data rates
Larger range
More users

4. A smart MIMO-OFDM system is key for 4G high-speed WLANs
The indoor propagation channel is best served by space/frequency processing
The MIMO hype: a critical review
A smart MIMO solution adapts to scene and user needs

5. The indoor channel suffers from severe multipath propagation
delay
Attenuation (dB)
delay dispersion
user
terminal
base
station
Need for ISI mitigation
angle dispersion
Need for full ST processing
DoA beamforming not suited

6. Indoor channel characterized by spatial and frequency selectivity
Angular response
Frequency response
Time response

7. MIMO-OFDM exploits spatial and frequency selectivity
MIMO Capacity: n-fold increase possible!
SISO
MIMO or SISO Capacity
(bit/s/Hz)
SNR (dB)
MIMO
2x2
3x3
4x4
1x1
Higher data rates
Larger range Less TX power
More users

8. SDM/SDMA enables higher data rates/more users
MIMO TX Proc. H
AP or UT
Stream 1
Stream 2
UT or AP
MIMO with TX pre-processing
Simple receiver
TX-CSI needed
Reciprocal transceiver at TX
MIMO-TX  SDMA-DL H
AP or UT
Stream 1
Stream 2
MIMO RX Proc.
UT or AP
MIMO with RX post-processing
No TX-CSI needed
Better for time- varying channels
More complex receiver (SIC,ML)
MIMO-RX  SDMA-UL
( Joint TX-RX processing possible )

9. Space-time coding enables larger range and/or less Tx power
MIMO Space-time encoding H
AP or UT
1 Stream
UT or AP
MIMO with Space-time (block) coding
MIMO Space-time Decoding
No TX-CSI needed
ML receiver with simple linear processing
No rate enhancement (rate 1 only for Nant = 2)
Also applies to any number of receive antennas
Space-time or space-frequency

10. Why be skeptical about MIMO?
Can we conquer the wireless MIMO channel?
Can we get the MIMO solution that meets the actual needs (rate/range, multi-user capacity, power)?
Can DSP complexity be mastered?
Can front-end cost and power be acceptable?
Can it elegantly bring its benefits in current systems and standards?

11. A smart MIMO system adapts to scene and actual user needs (1)
User HW profile
Optimal
Mode
Selection
User QoS
requirements
Channel
conditions
SDM, STBC, SDMA

12. A smart MIMO system adapts to scene and actual user needs (2)
Reference SISO case:
Pdc, mobile = 1, Rmax = 54 Mbps, Dist max = 1
SDM brings higher throughput in DL/UL
SDMA multiplies cell capacity
STBC brings robustness

13. SDMA multiplies cell capacity for single-antenna terminals
Reference SISO case:
Pdc, mobile = 1, Rmax = 54 Mbps, D max = 1
Downlink SDMA: Nant X Rmax
Uplink TDMA: PTx-dc, mobile =
ComplexityMobileTerm = SISO
ComplexityBasestation:
digital = Nant x SISO
analog = 0.9 Nant x SISO
Digital assuming:
for SISO digital 1/3 is in inner modem, 1/3 is in outer modem (FEC), 1/3 is in MAC
SDMA Tx processing only rises inner modem complexity
For 2 antennas SDMA Tx ~ 4.SISO
Assuming for analog the only ‘saving’ is in reuse of the LO
Pdc range is based on : lower limit (0.1) is based on ~ 10% of the channels is ‘really bad’ so antenna diversity saves on Tx power AND retransmissions (up to 5), higher limit (0.8) is based on 3dB gain and PowerAmp eats uo 40% of total transmit power

14. STBC brings increased robustness
Reference SISO case:
Pdc, mobile = 1, Rmax = 54 Mbps, D max = 1
STBC brings
Robustness almost for free
Assumptions on D:
Lower limit in case of line of sight using cross-polarized antennas, you get full diversity and P ~ D-2
Upper limit in case of one of these really bad channels
This does nowhere take into account retransmissions/goodput

15. SDM brings higher throughput
Reference SISO case:
Pdc, mobile = 1, Rmax = 54 Mbps, D max = 1
NT = NR = Nant
Throughputmax: Nant X Rmax, D = up to 2
ComplexityMobileTerm =
digital = 0.75 Nant x SISO
analog = 0.9 Nant x SISO
ComplexityBasestation:
digital = Nant x SISO
Assumptions on D:
Lower limit in case of line of sight using cross-polarized antennas, you get full diversity and P ~ D-2
Upper limit in case of one of these really bad channels
This does nowhere take into account retransmissions/goodput

16. Optimal use of flexibility in FEC in Rx
Where does the power go?
Standard SISO reference case for PRF = 40mW:
1 antenna, Tx
1 antenna, Rx
digital inner modem
digital outer modem
MAC
Converters
I/Q (de)modulator PA
Local oscillator
PDC=1.1
PDC=1.0
PDC=2.1
50%
Optimal use of flexibility in FEC in RX
early stop criterion
switching off in time of different modules
‘run-time’ optimal use of flexibility
Main messages
power distribution very different in TX and RX
everything you do at the TX to save power in PA greatly helps overall. MIMO can help here since it provides a better link budget
everything you do at the RX to reduce the FEC decoding effort greatly helps overall. MIMO can also help here since it has a better robustness
IMEC’s reference designs to Low Power:
Optimal use of flexibility in FEC in Rx
Doherty PA in Tx
PDC=0.7
PDC=0.7

17. How does power scale in MIMO?
Reference case: 2 parallel standard SISO at PRF = 40mW:
MT2 PDC=2.1
MT1 PDC=2.1
PDC=4.2
IMEC’s smart MIMO at PRF = 40mW:
AP PDC=2.6
MT PDC=2.1
50%
50%
PDC=2.3
Standard SISO case:
No Class A PA, no power gain in FEC
Ptx = 1610 mW Prx = 1450 mW
Reference MIMO case:
2 SISO case in parallel (at different frequencies) to get a capacity increase by 2,
no shared hw, no reuse
IMEC’s MIMO to show off:
All processing in AP (Tx proc in DL, Rx proc in UL), no processing in MT
Doherty in Tx with 25% power efficiency over the burst
Power gain in FEC in Rx of 10
Main messages
MIMO requires less power per transmitted bit: for the same rate almost reduced by half compared to reference case; so the general perception that MIMO is nice for high rate, but is complex and power hungry is not really justified
different distribution: more in DSP (technology scaling helps here) and convertors (!)
MIMO processing
digital outer modem
MAC
Converters
I/Q (de)modulator PA
Local oscillator
digital inner modem
PDC MT Tx
PDC MT Rx

18. Do MIMO upgrades ask for changes to 802.11a standard?
SDM s1 s2
TX-SDM
SDMA s1 s2
TX-SDMA
MRC s1
TX-MRC
STBC s1
STBC s1
stbc
Downlink
Uplink
SDM s1 s2
RX-SDM
SDMA s1 s2
RX-SDMA
MRC s1
RX-MRC
STBC s1
none
extra PHY mode
PHY+MAC

19. IMEC’s smart MIMO says YES to crucial questions
Can we get the MIMO solution that meets the actual needs (rate/range, multi-user capacity, power)?
Can DSP complexity be mastered?
Can front-end cost and power be acceptable?
Can it elegantly bring its benefits in current systems and standards?
Yes!