Lecture 15 Wireless Communication
Monsoon semester 2021
Topic: Direct-Sequence Spread-Spectrum Systems
Vinay Joseph
NIT Calicut
October 12, 2021
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Where are we in this course?
Module 1
▶ High level analysis of wireless communication networks using a
simplified model of wireless channel
▶ Careful modeling of a wireless channel
Module 2
▶ Point-to-point communication in wireless fading channel
Module 3
▶ Analysis of some wideband wireless communication and systems using
more detailed wireless channel model
⋆ Direct-sequence spread-spectrum communication and system
⋆ OFDM communication and system
▶ Main reference [1]: Chapter 3, Chapter 4 and Appendix A,
Fundamentals of Wireless Communication, Tse et al
⋆ Available at
https://web.stanford.edu/~dntse/wireless_book.html
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Narrowband vs wideband systems
Wireless systems can be broadly classified into
▶ Narrowband systems
▶ Wideband systems
Narrowband systems: user transmits using narrowband channels
▶ Narrowband channel’s bandwidth is smaller than channel coherence
bandwidth
▶ Narrowband channels can often be ”decoupled” using interferene
management (e.g., orthogonal resources per channel)
▶ Narrowband system thus can be analyzed by separate analysis of
individual channels
▶ Example: GSM (a 2G technology) uses this
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Wideband systems: Introduction
In wideband channels, user transmissions can span entire bandwidth
We focus on wideband systems in this module
▶ Wideband channels cannot be easily ”decoupled” since entire
bandwidth can be used by any users
▶ A system level view (in addition to channel-level view or link-level view)
is often useful for wideband systems
Examples
▶ Direct-sequence spread-spectrum systems
⋆ IS-95 (a 2G technology) uses CDMA which is based on direct-sequence
spread-spectrum
▶ Orthogonal Frequency Division Multiplexing (OFDM) based systems
⋆ 4G/LTE and 5G/NR use OFDM
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Wideband channel: Model
Wideband channel tends to
▶ be frequency-selective since transmission bandwidth is typically more
than channel coherence bandwidth
▶ have large delay spread compared to symbol duration
Wideband channel can be modeled using finite number L of taps:
L−1
X
y [m] = hl [m]x[m − l] + w [m]
l=0
▶ There is Inter-Symbol Interference (ISI)
▶ We saw a simple approach sending just one information symbol every L
slots. But, its data rate is low.
▶ Detection in wideband channel has to address ISI (directly or indirectly)
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Wideband systems: Detection approaches
PL−1
Wideband channel model: y [m] = l=0 hl [m]x[m − l] + w [m]
Three detection approaches for wideband channels
▶ Single-carrier systems with equalization
⋆ Single carrier is used with equalization (extracting transmitted symbols
from received signal with ISI) at receiver
⋆ Non-linear equalizers like ML based Viterbi algorithm perform well
though have high-complexity
⋆ Lower complexity linear equalizers (sub-optimal) can also be used
▶ Multi-carrier systems
⋆ Using (precoding) operations at transmitter, wideband ISI channel is
converted into orthogonal narrowband ”sub-carriers”
⋆ Frequency diversity achievable using sufficiently apart sub-carriers
⋆ Example: OFDM (considered in next lecture)
▶ Direct-sequence spread-spectrum (more on this soon)
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: Introduction
Information symbols modulated by a pseudonoise (PN) sequence and
transmitted over bandwidth W much larger than data rate R
How is ISI tackled? Because of low data rate, ISI is small across
(long) PN sequences
Downside: low rate for any one user
▶ Example: For IS-95 (CDMA), bandwidth is 1.2288MHz and a typical
data rate is just 9.6 kbits/s.
Then why use it? Multiple users can share a channel with users
appearing like noise to each other.
▶ Since ISI is low, simple receiver possible without needing equalization
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: Plan for this lecture
Using direct-sequence spread-spectrum communication for a
wideband channel:
▶ Understand how transmission works and role of PN sequence
▶ How to perform detection?
⋆ We see how ”Rake Receiver” can be used for detection.
▶ What performance do we get?
⋆ We see that we can get diversity gain of L
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: Setting
Consider frequency selective channel with finite number L of taps:
L−1
X
y [m] = hl [m]x[m − l] + w [m]
l=0
▶ Assume hl [m] ∼ CN (0, 1/L) (i.e., equal energy per tap).
▶ Usually, there is more energy in taps associated with shorter delays
Binary information symbol is transmitted by transmitting one of two
n-chips long PN sequences xA or xB
▶ Shifted versions of a PN sequence are almost orthogonal (more precise
definition soon)
▶ One bit is sent every n/W seconds (chip duration ≈ 1/W ). So,
n = W /R
▶ W /R is also called processing gain
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: Setting
Assume hl [m] does not vary with m during transmission of PN
sequence (i.e., over n chips’ duration)
▶ That is, Wn << Coherence-time
▶ So, we can consider following model:
L−1
X
y [m] = hl x[m − l] + w [m]
l=0
n
We also assume that W >> delay-spread, so that there is no
significant ISI across PN sequences
▶ Can thus focus on detection of one PN sequence at a time
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: Transmission
We focus on antipodal signaling case: x A = u , x B = −u , where u is
a PN sequence
Notation
▶ u is n dimensional vector u = [u[1], ..., u[n]]
▶ u (l) is n + l dimensional vector obtained by shifting u by appending l
zeros before u and appending L − l zeros after u
⋆ i.e., u (l) = [0, ..., 0, u[1], ..., u[n], 0, ..., 0]
Channel model can be rewritten considering one PN sequence at a
time as
L−1
y= hl x (l) + w
X
l=0
where x (l) is either u (l) or −u (l) depending on whether x A or x B is
sent respectively
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Direct-sequence spread-spectrum: PN sequence property
Use of PN sequence is a distinguishing feature of spread spectrum
systems
Key property of PN sequence: Shifted versions of a PN sequence
are nearly orthogonal to each other. So, if u is a pseudonoise
sequence, then for any l = 0, ..., L − 1
∗ n
(l ′ )
u u
X
(l)
<< |u[i]|2 , l ̸= l ′
i=1
Note that
∗ ′
u (l) u (l ) = ∥u ∥2 , l = l ′
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
How to detect at receiver? Exploit PN sequences property
Channel model: y = L−1 l=0 hl x
(l) + w , x (l) is either u (l) or −u (l)
P
Consider following sufficient statistics for each l = 0, ..., L − 1:
∗ ∗ L−1
u (l) u (l)
!
(l ′ )
y= hl ′ x + w
X
(l)
r =
∥u ∥ ∥u ∥ ′
l =0
∗ (l ′ ) ∗
L−1
X u (l) x u (l) w
= hl ′ +
′
∥u ∥ ∥u ∥
l =0
u is pseudonoise sequence, i.e.,
∗ n
(l ′ )
u u
X
(l)
<< |u[i]|2 , l ̸= l ′
i=1
∗
(l ′ )
u (l)
u = ∥u ∥2 , l = l ′
Hence, r (l) = hl x + w (l) , l = 0, ..., L − 1, where x ± ∥u ∥
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
How to detect at receiver? End with an MRC
Following is reminiscent of channel model with repetition coding:
r (l) = hl x + w (l) , l = 0, ..., L − 1, where x ± ∥u ∥
=⇒ r = h x + w , where x ± ∥u ∥
▶ Note that |x| = ∥u ∥ ∝ n = Processing-gain
⋆ This justifies use of the phrase ”processing gain”
▶ Note that w (l) ∼ CN (0, N0 ) for each l = 0, ..., L − 1, since
(l) ∗
u w
w (l) =
∥u ∥
Like in repetition coding, MRC operation can be used to decide
whether x is ∥u ∥ or −∥u ∥:
∗ x=∥u ∥
h
R r ≥ 0
∥h ∥ <
x=−∥u ∥
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Rake receiver
Receiver also called Rake receiver performs following operations using
received signal y and using knowledge of (estimate of) h
(u (l) )
∗
(i) Compute r (l) = ∥u ∥ y , for each l = 0, ..., L − 1
(ii) Compute R ∥hh ∥ r
n ∗ o
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Rake receiver: Performance
Using analysis similar to repetition coding with MRC, we have
s
2∥ h ∥2
∥ u ∥2
pe = E [P {error|h }] = E Q
N0
" !#
∥h ∥2 ∥u ∥2
≤ E exp −
N0
1
≤
1 + ∥Nu0∥L
2
L
▶ Last step uses a variation of analysis of repetition coding with MRC to
account for variance 1/L of hl [m] ∼ CN (0, 1/L)
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Rake receiver: Diversity gain
Probability of error upper bound with Rake receiver:
1
pe ≤
1 + ∥Nu0∥L
2
L
Rake receiver provides diversity gain of L
However ∥Nu0∥L decreases with L and this can be attributed to splitting
2
of energy across L taps
▶ Thus, decay of pe with L is slower
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
DSSS communication for wideband channel: Summary
DSSS communication can be used in wideband systems
Transmission uses PN sequence:
L−1
y= hl x (l) + w
X
l=0
where x (l) is either u (l) or −u (l) , where u is a PN sequence
Detection uses Rake receiver:
(u (l) )
∗
i) Compute r (l) = ∥u ∥ y , for each l = 0, ..., L − 1
ii) Detection rule
∗ x=∥u ∥
h
R r ≥ 0
∥h ∥ <
x=−∥u ∥
1
Rake receiver provides diversity gain of L: pe ≤ L
∥u ∥2
1+ N L
0
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication
Reference
[1] Tse, D., Viswanath, P. (2005). Fundamentals of Wireless
Communication. Cambridge: Cambridge University Press
Vinay Joseph (NIT Calicut) Lecture 15 Wireless Communication