TSKS14
Multiple Antenna
Communications
Lecture 2, 2020
Emil Björnson
TSKS14 Multiple Antenna Communications 2020-03-30 2
Taxonomy: Point-to-point channels
Single-input single-output (SISO) Multiple-input single-output (MISO)
Single-input multiple-output (SIMO) Multiple-input multiple-output (MIMO)
TSKS14 Multiple Antenna Communications 2020-03-30 3
Outline of this lecture
• Recap of SISO channel capacity
• Capacity of SIMO and MISO channels
• Channel models for line-of-sight channels
• Beamwidth with uniform linear arrays
TSKS14 Multiple Antenna Communications 2020-03-30 4
Not all transmit power is received
• Wireless signals spread out as a sphere
"#
• Fraction of received signal at distance 𝑑: $%& #
• Cellular communication: 70 dB to 130 dB is lost! Also known as
• Channel gain 𝑔 , is around −70 dB to −130 dB pathloss
𝑛[𝑙]
𝑥[𝑙] 𝑔 + 𝑦 𝑙 = 𝑔 ⋅ 𝑥 𝑙 + 𝑛[𝑙]
Channel response
TSKS14 Multiple Antenna Communications 2020-03-30 5
Capacity of memoryless SISO channel
• AWGN channel with a complex channel gain 𝑔:
𝑦 𝑙 = 𝑔 ⋅ 𝑥 𝑙 + 𝑛[𝑙]
• 𝑥 𝑙 ∼ 𝐶𝑁(0, 𝑞), energy per sample: 𝑞 = 𝑃/𝐵
• 𝑛 𝑙 ∼ 𝐶𝑁(0, 𝑁A)
Memoryless channel: Drop dependence on 𝑙
• Capacity computation: Treat 𝑔 ⋅ 𝑥 ∼ 𝐶𝑁(0, 𝑞 𝑔 ,) as the signal!
𝑞𝑔,
C = log , 1 + bits per complex sample
𝑁A
How can we increase the SNR?
Received SNR
TSKS14 Multiple Antenna Communications 2020-03-30 6
Single-input multiple-output (SIMO) channel
𝑛Q
𝑦Q
+
𝑔Q 𝑦,
𝑥 𝑔, +
Noise: 𝑛W ∼ 𝐶𝑁(0, 𝑁A )
𝑔S 𝑛,
+ 𝑦S
𝑛S
• Received vector signal:
𝑦Q 𝑔Q 𝑛Q
⋮ = ⋮ 𝑥+ ⋮
𝑦S 𝑔S 𝑛S
• Short form:
𝒚 = 𝒈𝑥 + 𝒏
TSKS14 Multiple Antenna Communications 2020-03-30 7
Geometry of received signal
• Received vector 𝒚 is summation of
• Desired signal 𝒈𝑥
• Noise vector 𝒏
• Want to turn 𝒚 into an
estimate of 𝑥
Only part of 𝒚 parallel to 𝒈
contains desired signal
TSKS14 Multiple Antenna Communications 2020-03-30 8
Capacity of SIMO channel
• Maximum ratio combining (MRC):
𝒈X 𝒈X 𝒏
𝒚= 𝒈 𝑥+
𝒈 𝒈
Combining vector Scalar
Like a SISO channel with 𝑔 = 𝒈
• Capacity:
𝑞 𝒈 ,
C = log , 1+ bits per complex sample
𝑁A
TSKS14 Multiple Antenna Communications 2020-03-30 9
Multiple-input single-output (MISO) channel
𝑥Q 𝑔Q 𝑛 ∼ 𝐶𝑁(0, 𝑁A )
𝑥,
𝑔,
+ 𝑦
𝑥S
𝑔S
• Received vector signal:
𝑥Q
𝑦 = 𝑔Q ⋯ 𝑔S ⋮ + 𝑛 = 𝒈Z 𝒙 + 𝑛
𝑥S
• Precoding: 𝒙 = 𝒘𝑥]
Information signal: 𝑥] ∼ 𝐶𝑁(0, 𝑞)
Unit-norm
precoding vector
TSKS14 Multiple Antenna Communications 2020-03-30 10
Geometry of received signal
• Channel projects signal onto 𝒈∗ :
Maximum ratio transmission (MRT):
𝒈∗
𝒘=
𝒈
𝒈∗
𝑦 = 𝒈Z 𝑥] + 𝑛 = 𝒈 𝑥] + 𝑛
𝒈
• Capacity:
𝑞 𝒈 ,
C = log , 1+ bits per complex sample
𝑁A
TSKS14 Multiple Antenna Communications 2020-03-30 11
Receive and transmit beamforming
• Same capacity of SIMO and MISO channel
𝑞 𝒈 ,
C = log , 1 + bits per complex sample
𝑁A
• Achieved by “beamforming” along channel vector 𝒈
• Beamforming gain of 𝒈 , (often proportional to 𝑀)
• Several different names
• Maximum ratio combining/transmission (MRC, MRT)
• Conjugate beamforming, matched filtering
TSKS14 Multiple Antenna Communications 2020-03-30 12
Transmit beamforming
Directivity → Beamforming gain
• Other names:
• Array gain
• Power gain
TSKS14 Multiple Antenna Communications 2020-03-30 13
Free-space line-of-sight commmunication
• Direct path between transmitter and receiver
• No other paths
• We will derive a channel model
TSKS14 Multiple Antenna Communications 2020-03-30 14
Complex baseband representation
• Real-valued baseband signals (bandwidth 𝐵/2):
𝑠c (𝑡) and 𝑠e (𝑡)
• Complex baseband representation:
𝑠f 𝑡 = 𝑠c 𝑡 + 𝑗𝑠e (𝑡)
• Real-valued passband signal
𝑠 𝑡 = 𝑠c 𝑡 2 cos 2𝜋𝑓j 𝑡 − 𝑠e 𝑡 2 sin 2𝜋𝑓j 𝑡
= 𝑅𝑒 2𝑠f 𝑡 𝑒 n,%op q
• Carrier frequency: 𝑓j ≫ 𝐵
• Wavelength: 𝜆 = 𝑐/𝑓j
TSKS14 Multiple Antenna Communications 2020-03-30 15
Complex baseband representation (2)
Fourier transforms: 𝑆f 𝑓 = ℱ 𝑠f 𝑡
1
𝑆 𝑓 = 𝑆f 𝑓 − 𝑓j + 𝑆f∗ −𝑓 − 𝑓j
2
𝑆f 𝑓
𝐵/2
𝑆 𝑓 𝐵
−𝑓j +𝑓j
TSKS14 Multiple Antenna Communications 2020-03-30 16
Uniform linear array (ULA)
Antenna spacing: Δ
Angle of arrival: 𝜑
TSKS14 Multiple Antenna Communications 2020-03-30 17
Impact of channel in complex baseband
• Transmitted signal:
𝑠 𝑡 = 𝑅𝑒 2𝑠f 𝑡 𝑒 n,%op q
• Received signal at distance 𝑑W :
𝜆 𝑑W 𝜆 𝑑W n,%op &
q{ j|
𝜇W 𝑡 = 𝑠 𝑡− = 𝑅𝑒 2𝑠f 𝑡− 𝑒
4𝜋𝑑W 𝑐 4𝜋𝑑W 𝑐
• Receiver can compensate for time delay by selecting sampling time
• Different 𝑑W : Only one antenna has perfect sampling
TSKS14 Multiple Antenna Communications 2020-03-30 18
Far-field approximation
• Distances 𝑑Q, … , 𝑑S approximately the same
Approximately same
channel gain:
𝜆, 𝜆,
, ≈
4𝜋𝑑W 4𝜋𝑑 ,
𝑑W n,%op &
q{ j|
𝑅𝑒 2𝑠f 𝑡− 𝑒
𝑐
&• &
Sample at 𝑡 = =
j j
TSKS14 Multiple Antenna Communications 2020-03-30 19
ULA channel model
• 𝑀 antennas in receiver array
• Angle of arrival 𝜑
• Wavelength 𝜆
• Distance Δ between antennas
• Distance 𝑑 to first antenna
• Channel vector:
1
€ •‚ƒ „
𝜆 𝑒
{n,%
" Applies to both SIMO and
𝒈=
4𝜋𝑑 ⋮ MISO channels
S{Q € •‚ƒ „
{n,%
𝑒 "
TSKS14 Multiple Antenna Communications 2020-03-30 20
Channel capacity with line-of-sight channel
• Recall:
𝑞 𝒈 ,
C = log , 1+ bits per complex sample
𝑁A
, "#
• We have 𝒈 = 𝛽𝑀 where 𝛽 = $%& # :
𝑞𝛽𝑀
C = log , 1+
𝑁A
Beamforming gain: 𝑀
TSKS14 Multiple Antenna Communications 2020-03-30 21
Beamwidth
• Beamforming gain in more
than one direction
• Example: 𝑀 = 10, Δ = 𝜆/2
• First-null beamwidth:
4
≈
𝑀
More antennas:
Narrower beams
TSKS14 Multiple Antenna Communications 2020-03-30 22
Two benefits of
beamforming
• Beamforming gain
• Stronger signal at
desired location
• Narrower beam
• Less interference at
undesired locations
End of Lecture 2
TSKS14 Multiple Antenna
Communications